Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text...

24
Group 6 metal complexes as electrocatalysts of CO2 reduction: strong substituent control of the reduction path of [Mo( 3-allyl)(CO)2(x,x -dimethyl-2,2 - η bipyridine)(NCS)] (x = 4–6) Article Accepted Version Taylor, J. O., Veenstra, F. L. P., Chippindale, A. M., Calhorda, M. J. and Hartl, F. (2019) Group 6 metal complexes as electrocatalysts of CO2 reduction: strong substituent control of the reduction path of [Mo( 3-allyl)(CO)2(x,x -dimethyl-2,2 - η bipyridine)(NCS)] (x = 4–6). Organometallics, 38 (6). pp. 1372- 1390. ISSN 1520-6041 doi: https://doi.org/10.1021/acs.organomet.8b00676 Available at http://centaur.reading.ac.uk/81186/ It is advisable to refer to the publisher’s version if you intend to cite from the work.  See Guidance on citing  . To link to this article DOI: http://dx.doi.org/10.1021/acs.organomet.8b00676 Publisher: American Chemical Society All outputs in CentAUR are protected by Intellectual Property Rights law, 

Transcript of Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text...

Page 1: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

Group 6 metal complexes as electrocatalysts of CO2 reduction: strong substituent control of the reduction path of  [Mo( 3­allyl)(CO)2(x,x ­dimethyl­2,2 ­η ′ ′bipyridine)(NCS)] (x = 4–6) Article 

Accepted Version 

Taylor, J. O., Veenstra, F. L. P., Chippindale, A. M., Calhorda, M. J. and Hartl, F. (2019) Group 6 metal complexes as electrocatalysts of CO2 reduction: strong substituent control of the reduction path of [Mo( 3­allyl)(CO)2(x,x ­dimethyl­2,2 ­η ′ ′bipyridine)(NCS)] (x = 4–6). Organometallics, 38 (6). pp. 1372­1390. ISSN 1520­6041 doi: https://doi.org/10.1021/acs.organomet.8b00676 Available at http://centaur.reading.ac.uk/81186/ 

It is advisable to refer to the publisher’s version if you intend to cite from the work.  See Guidance on citing  .

To link to this article DOI: http://dx.doi.org/10.1021/acs.organomet.8b00676 

Publisher: American Chemical Society 

All outputs in CentAUR are protected by Intellectual Property Rights law, 

Page 2: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

including copyright law. Copyright and IPR is retained by the creators or other copyright holders. Terms and conditions for use of this material are defined in the End User Agreement  . 

www.reading.ac.uk/centaur   

CentAUR 

Central Archive at the University of Reading 

Reading’s research outputs online

Page 3: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

1

Group-6 Complexes as Electrocatalysts of CO2 Reduction: Strong

Substituent Control of the Reduction Path of [Mo(η3-allyl)(CO)2(x,x′-

dimethyl-2,2′-bipyridine)(NCS)] (x = 4-6)

James O. Taylor,1 Florentine L. P. Veenstra,1,2 Ann M. Chippindale,1 Maria José Calhorda,3 František Hartl1,*

1Department of Chemistry, University of Reading, Reading, RG6 6AD, United Kingdom. 2Institute for Chemical and Bioengineering, Department of Chemistry and Applied Biosciences, ETH Hönggerberg, Vla-dimir-Prelog-Weg 1, 8093 Zurich, Switzerland 3 Centro de Química e Bioquímica and BioISI - Biosystems & Integrative Sciences Institute, Departamento de Quı́mica e Bioquı́mica, Faculdade de Ciências, Universidade de Lisboa, 1749-016 Lisboa, Portugal

ABSTRACT: A series of complexes [Mo(η3-allyl)(CO)2)(x,x′-dmbipy)(NCS)] (dmbipy = dimethyl-2,2ʹ-bipyridine; x = 4-6) have been synthesized and their electrochemical reduction investigated using combined cyclic voltammetry (CV) and variable-tempera-ture spectroelectrochemistry (IR/UV-vis SEC) in tetrahydrofuran (THF) and butyronitrile (PrCN), at gold and platinum electrodes. The experimental results, strongly supported by DFT calculations, indicate that the general cathodic path of these Group-6 organo-metallic complexes is closely related to that of the intensively studied class of Mn tricarbonyl α-diimine complexes, themselves re-cently identified as important smart materials for catalytic CO2 reduction. The dimethyl substitution on the 2,2ʹ-bipyridine ligand backbone has presented new insights into this emerging class of catalysts. For the first time, the 2e‒ reduced 5-coordinate anions [Mo(η3-allyl)(CO)2)(x,x′-dmbipy)]‒ were directly observed with IR SEC. The role of steric and electronic effects in determining the reduction-induced reactivity was also investigated. For the 6,6′-dmbipy, the primary 1e‒ reduced radical anions exert unusual sta-bility radically changing the follow up cathodic path. The 5-coordinate anion [Mo(η3-allyl)(CO)2)(6,6′-dmbipy)]‒ remains stable at low temperature in strongly coordinating butyronitrile and does not undergo dimerization at elevated temperature, in sharp con-trast to reactive [Mo(η3-allyl)(CO)2)(4,4′-dmbipy)]‒ that tends to dimerize in a reaction with the parent complex. The complex with the

5,5′-dmbipy ligand combines both types of reactivity. Under aprotic conditions, the different properties of [Mo(η3-allyl)(CO)2)(x,x′-dmbipy)]‒ are also reflected in their reactivity towards CO2. Preliminary CV and IR SEC results reveal differences in the strength of CO2 coordination at the free axial position. Catalytic waves attributed to the generation of the 5-coordinate anions were observed by CV, but only a modest catalytic performance towards the production of formate was demonstrated by IR SEC. For 6,6′-dmbipy, a stronger catalytic effect was observed for the Au cathode compared to Pt.

INTRODUCTION

Atmospheric levels of anthropogenic CO2 have continued to increase unabated for several years, driving a concurrent rise in yearly average temperatures and extreme weather pat-terns.1 Although it has an obvious damaging impact on the en-vironment, CO2 itself has potential to become a sustainable source of carbon-based fuels and chemical feedstocks for Fischer-Tropsch chemistry. To fulfill this potential, CO2 must

first be reduced electro- or photochemically, converting it from a harmful by-product of industry to compounds with the afore-mentioned applications.2,3 However, the direct electrochemical reduction to CO2

•‒ is hindered by large overpotentials and one must take advantage of proton-coupled reduction pathways.3,4 An exception is a dinuclear Cu(I) catalyst reported recently by Bowman and co-workers, which reduces CO2 from the air to ox-

Page 4: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

alate isolated as a lithium salt.5 In comparison, the proton-cou-pled paths have significantly reduced energy barriers to catal-ysis but they must be promoted by homogenous or hetero-genous (Cu) transition-metal catalysts. The effect of the cata-lyst is two-fold, enabling reduction of CO2 at much lower over-potentials whilst also allowing one to handle a larger number of proton- and electron-transfer steps facilitating the transfor-mation to more complex and valuable reduction products such as formaldehyde, methanol or methane.6 Of the many homoge-neous transition-metal electrocatalysts of CO2 reduction known in the literature, the most widely studied have been those based on the noble metals, Re7–12, Ru10,13–15, Rh16 and Ir.17,18 Although first reported almost 40 years ago, they are still offering up both interesting redox properties and syn-thetic challenges. The high costs of these metals, as well as a strong competition for the limited supply from the electronics industry, largely preclude the possibility of any industrial-scale system. Thus, efforts now focus on the more abundant metals, Mn, Fe, Co19, and Ni20,21. Highly promising are the sys-tems of [Mn(CO)3)(α-diimine)X] (X = (pseudo)halide), only re-cently found to be catalytically active at potentials similar to the analogous complexes of rhenium, in the presence of Brønsted or Lewis acids.22–28 The Fe-porphyrin systems are also highly efficient catalysts of CO2 reduction, but syntheti-cally challenging to prepare.29–31

Surprisingly, and despite the analogues found in Nature,

the Group-6 metal triad (Cr, Mo, W) has largely been ignored by the wider community. This has been highlighted in several recent comprehensive reviews of the field.32–34 Preliminary studies of Group-6 complexes show a potential to catalytically reduce CO2 with a comparable activity and efficiency to that of the widely studied Group-7 systems. Recently, it was reported that Group-6 hexacarbonyl complexes can behave catalytically toward CO2.35 This is highly unusual, as most catalysts reported to date bear a redox-active ligand acting as both an electron reservoir and a parking place for the protons required for ca-talysis. The hexacarbonyl is reduced by 2e‒ to form [M(CO)5]2‒ that can bind CO2 and transform to [M(CO)5(CO2)]2‒; under an-hydrous conditions, the major product of the reduction is CO. Interestingly, addition of water or another proton source in-hibits the activity of the catalyst, most likely due to competing formation of H2 from [M(CO)5H]‒. A comparison with [M(CO)5]‒ (M = Re or Mn), formed by the reduction of the cor-responding decacarbonyl dimers, shows that the Group-7 met-als are not active toward CO2 in this manner. Instead, they simply bind CO2 and require a non-innocent ligand to trigger the reduction. The difference in their reactivity may simply be explained by the fact that the Group-6 [M(CO)5]2‒ species are better reducing agents, due to the negative charge experienc-ing less nuclear charge than the comparable Group-7 species. Apart from this, the limited literature regarding Group-6 cata-lysts has mostly focused on the family of [M(CO)4(α-diimine)] catalysts.36–40 The first of these studied as a catalyst of CO2 re-duction was [Mo(CO)4(bipy)] (bipy = 2,2ʹ-bipyridine).36,37 The parent complex is reduced in two consecutive 1e‒ steps to form first the corresponding radical anion and then the catalytically active 5-coordinate species, [Mo(CO3(bipy)]2‒. The latter is iso-electronic with the Group-7 catalysts, [M(CO)3)(bipy)]‒ (M = Re, Mn), formed by 2e‒ (ECE) reduction of [M(CO)3(bipy)X] (X = (pseudo)halide).

Unlike their Group-7 cousins, [Mo(CO)4(α-diimine)] (α-

diimine = 2,2ʹ-bipyridine (bipy) and x,xʹ-dimethyl-bipyridine

(dmbipy, x = 4-6) show higher sensitivity to the cathodic ma-terial.37 This can be exploited in the form of a low-energy path-way only accessible with the use of a gold cathode that facili-tates CO dissociation from the usually stable radical anion, [Mo(CO)4(bipy)]•‒, converting it to [Mo(CO)3(bipy)]•‒. The lat-ter 5-coordinate complex is reducible to the dianionic catalyst at a cathodic potential only slightly more negative than the in-itial reduction of the neutral tetracarbonyl parent, thereby by-passing the energetically demanding formation of the unstable dianion, [Mo(CO)4(bipy)]2‒. This CO dissociation-first pathway is similar to the low-energy protonation-first pathway re-ported for [Mn(CO)3(bipy)Br].41 These observations have been confirmed by Cowan and co-workers using sum frequency gen-eration (SFG) spectroscopy.42 One can also alter the pathway these Group-6 tetracarbonyl catalysts follow, by substitution on the 2,2′-bipyridine rings.39 Advantageous positioning of ste-ric bulk at the 6,6′-position can enhance the CO dissociation from the primary radical anion, improving the catalytic activ-ity, while substitution at the 4,4'-position suppresses the CO dissociation, forcing the catalyst to follow the higher-energy reduction-first pathway. Smart choice of the solvent and elec-trolyte, especially turning to N-methyl-pyrrolidone (NMP), can improve the catalytic efficiency remarkably. Nervi and co-workers38 have described a comparable catalytic action for [M(CO)4(α-diimine)] (M = Mo, W; α-diimine = 4,6-diphenyl-bipy, 4-dimethoxyphenyl-6-phenyl-bipy). A higher catalytic ac-tivity towards conversion of CO2 to CO and a unique activation pathway have also been reported for the complexes [M(CO)4(2,2'-dipyridylamine)] (M = Mo and W), in which the catalytic center resides at the bidentate ligand rather than at the metal.38 The unusual cathodic behaviour of [W(CO)4(dpa)] has been rationalized with the help of quantum mechanical cal-culations. The 1e‒ reduction of the parent complex induces a rearrangement of the dpa ligand with one of the pyridine rings rotating away from the metal centre, which facilitates the re-duction of CO2.43

Inspired by the promising performance of the Group-6

tetracarbonyl catalysts, the related complex [Mo(η3-al-lyl)(CO)2(bipy)X] (X = (pseudo)halide) was identified as a pre-cursor to the catalytically-active 5-coordinate anion, [Mo(η3-allyl)(CO)2(bipy)]‒.44,45 In an analogous fashion to the well-studied catalyst precursor [Mn(CO)3(bipy)Br], the radical an-ion [Mo(η3-allyl)(CO)2(bipy)(NCS)]•‒, formed by the initial one-electron reduction (Epc = -1.99 V vs Fc/Fc+) is very unsta-ble at room temperature. The concomitant dissociation of the NCS‒ ligand yields the 5-coordinate radical [Mo(η3-al-lyl)(CO)2((bipy)] that can be reduced at the applied electrode potential of the parent complex to [Mo(η3-allyl)(CO)2(bipy)]‒. This anion reacts with as yet non-reduced parent, forming the dimer [Mo(η3-allyl)(CO)2(bipy)]2, the final step of the ECEC pathway. Here the similarity with the Mn-catalysts appears to end. In the original paper45 it was claimed that the 5-coordi-nate anion had been detected by IR spectroelectrochemistry; however, this assignment was only tentative. The main issue to be clarified in this work is under exactly what conditions an IR-detectable quantity of the 5-coordinate anionic catalyst can be formed. Indeed, the dimer-like species produced in this in-stance appear to be more reactive than the related hexacar-bonyl Mn‒Mn bound dimers and can only be detected on the relatively short time scale of cyclic voltammetry, The subse-quent reduction of the Mo‒Mo dimer in the presence of CO2 is then associated with catalytic current, and infrared monitoring at the catalytic wave reveals the formation of CO and free for-mate.

Page 5: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

Hereinafter, the study of the original (allyl)Mo‒bipy com-

plex is extended to the series of three x,xʹ-dmbipy derivatives (Chart 1), with the following goals: (i) to identify (by IR spec-troelectrochemistry) the 5-coordinate anions [Mo(η3-al-lyl)(CO)2(x,x′-dmbipy)]‒, the proposed active catalysts in these systems, which were misassigned for the reference bipy com-plex; (ii) to assess the catalytic performance of the 5-coordi-nate anions via CV and IR SEC; (iii) to investigate the steric and electronic consequences of the dimethyl substitution at the x = x′ = 4, 5 and 6 positions of the 2,2′-bipyridine ligand on the ca-thodic paths; (iv) to probe the variation of the cathodic mate-rial, Pt and Au, and its impact on the reduction potentials, ca-thodic pathways and catalytic performance, in analogy with [Mo(CO)4-n(x,x′-dmbipy)]•‒ (n = 0, 1; x = 4-6).

Chart 1. General molecular structure of the studied complexes [Mo(η3-allyl)(CO)2(x,x′-dmbipy)(NCS)] 1 (x = 4), 2 (x = 5) and 3 (x = 6).

EXPERIMENTAL SECTION

Materials and Methods. All syntheses and electrochemical meas-urements were conducted under a strictly inert atmosphere of dry argon, using standard Schlenk techniques. Solvents were freshly distilled under N2 from a mixture of Na/benzophenone (tetrahy-drofuran, THF), P2O5 (acetonitrile, MeCN) or CaH2 (butyronitrile, PrCN, and dichloromethane, DCM). The electrolyte, tetrabu-tylammonium hexafluorophosphate (TBAH, Agro-Organics), was recrystallized twice from hot ethanol and dried under vacuum at 80 °C. All other reagents were purchased from Sigma-Aldrich and used as received. The precursor [Mo(η3-allyl)(CO)2(MeCN)2(NCS)] was prepared according to the literature procedure.46 All the [Mo(η3-allyl)(CO)2(x,x′-dmbipy)(NCS)] complexes were prepared by facile thermal substitution of the labile acetonitrile ligands in [Mo(η3-allyl)CO)2(MeCN)2(NCS)]. For the electrocatalytic studies, solvents were saturated with CO2 (BOC, < 99.99%) at an atmos-pheric pressure by short bubbling on a frit.

General Synthesis of [Mo(η3-allyl)(CO)2(x,x′-dmbipy)(NCS)]. Under argon, a three-necked round-bottom flask was charged with [Mo(η3-allyl)CO)2(MeCN)2(NCS)] (230 mg, 0.69 mmol) and the ap-propriate x,x′-dimethyl-2,2′-bipyridine ligand (x = 4-6) in a small excess (typically 0.78 mmol). After the addition of dry deaerated DCM (10 mL), the mixture was refluxed at 50 °C for 2 h. The solu-tion was then allowed to cool to ambient temperature and its vol-ume was reduced by half. The remaining solution was transferred to a Schlenk vessel, followed by addition of hexane (5 x 10 mL) to precipitate the crude product that was further purified by extrac-tion with hexane and column chromatography on silica, using di-chloromethane:hexane 10:1 (v/v) as eluent. The yields of a red mi-crocrystalline powder varied between 35-55%. X-ray quality crys-tals were grown by vapour diffusion in hexane/DCM.

[Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(NCS)] (1). The precursor com-plex [Mo(η3-allyl)(CO)2(MeCN)2(NCS)] (230 mg, 0.69 mmol) and

4,4ʹ-dmbipy (151 mg, 0.82 mmol) reacted to afford 1 (168 mg, 55 %). 1H NMR (400 MHz, CD2Cl2): δ 8.55 (2H, d, J = 6 Hz), 7.85 (2H, s), 7.27 (2H, d, J = 5.6 Hz), 3.08 (2H, d, J = 6.4 Hz), 2.93 (1H, m), 2.54 (6H, s), 1.34 (2H, d, J = 6 Hz); (400 MHz, (CD3)2SO) 8.67 (2H, d, J = 5.2 Hz), 8.48 (2H, s), 7.53 (2H, d, J =5.6 Hz), 3.26 (2H, d, J = 6.8 Hz ), 3.19 (1H, m), 2.52 (6H, s), 1.42 (2H, d, J = 9.2 Hz). IR (THF): ν(CO) 1949, 1869 cm-1, ν(CN) of NCS‒ 2076 cm-1.

[Mo(η3-allyl)(CO)2(5,5ʹ-dmbipy)(NCS)] (2). The precursor com-plex [Mo(η3-allyl)(CO)2(MeCN)2(NCS)] (220 mg, 0.66 mmol) and 5,5ʹ-dmbipy (150 mg, 0.82 mmol) reacted to afford 2 (132 mg, 43 %). 1H NMR (400 MHz, CD2Cl2): δ 8.52 (2H, s), 7.87 (2H, d, J = 8 Hz), 7.75 (2H, d, J = 8 Hz), 3.12 (2H, d, J = 8 Hz), 2.96 (1H, m), 2.40 (6H, s), 1.36 (2H, d, J = 8 Hz); (400 MHz, (CD3)2SO) δ 8.64 (2H, s), 8.42 (2H, d, J =10 Hz), 8.05 (2H, d, J = 10 Hz), 3.40 (2H, d, J = 8 Hz), 3.19 (1H, m), 2.48 (6H, s), 1.40 (2H, d, J = 8 Hz). IR (THF): ν(CO) 1951, 1871 cm-1, ν(CN) of NCS‒ 2076 cm-1. [Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)(NCS)] (3). The precursor [Mo(η3-allyl)(CO)2(MeCN)2(NCS)] (230 mg, 0.69 mmol) and 6,6ʹ-dmbipy (143 mg, 0.78 mmol) reacted to afford 3 (109 mg, 36 %). 1H NMR (400 MHz, CD2Cl2): δ 7.82 (3H, m), 7.38 (2H, d, J = 8.4 Hz), 7.24 (1H, s), 2.94 (6H, s), 2.62 (3H, m), 1.19 (2H, d, J = 8 Hz); (400 MHz, (CD3)2SO): δ 8.45 (2H, d, J = 8 Hz), 8.18 (2H, t, J = 24 Hz), 7.72 (2H, d, J = 8 Hz), 2.99 (6H, s), 2.90 (1H, m), 2.80 (2H, d, J = 8 Hz), 1.30 (2H, d, J = 8 Hz). IR (THF): ν(CO) 1948, 1866 cm-1; ν(CN) of NCS‒ 2074 cm-1.

X-ray Structure Determination. A crystal of [Mo(η3-al-lyl)(CO)2(x,x′-dmbipy)(NCS)] was quickly mounted under Para-tone-N oil and flash-cooled to 150 K in a stream of nitrogen in an Oxford Cryostream cooler. Single-crystal X-ray intensity data were collected using an Agilent Gemini S Ultra diffractometer (Mo Kα ra-diation (λ = 0.71073 Å)). The data were reduced within the CrysA-lisPro software.47 The structure was solved using the program Su-perflip48, and all non-hydrogen atoms were located. Least-squares refinements were carried out using the CRYSTALS suite of pro-grams.49 The non-hydrogen atoms were refined anisotropically. Each hydrogen atom of the ligands was placed geometrically with a C‒H distance of 0.95 Å and a Uiso of 1.2-1.5 times the value of Uiso of the parent C atom. The positions of the hydrogen atoms were then refined with riding constraints. In the case of [Mo(η3-al-lyl)(CO)2(6,6ʹ-dmbipy)(NCS)], several disordered solvent mole-cules (DCM), which could not be modelled satisfactorily, were also present within the structure. The PLATON SQUEEZE software ena-bled the contribution to diffraction of the four disordered solvent molecules to be calculated, and thus it was possible to produce sol-vent-free diffraction intensities.50 CCDC codes: 1 – 1862262; 2 – 1862263; 3 – 1862264.

Cyclic Voltammetry. Cyclic voltammograms were recorded on a PGSTAT 302N potentiostat (Metrohm Autolab) under an atmos-phere of argon or carbon dioxide, using an air-tight three-electrode single-compartment cell. A Pt (128 μm) or Au (130 μm) disc micro-electrode polished by a 0.25-μm diamond paste (Sommer) served as the working electrode; the counter and pseudo-reference elec-trodes were made of coiled Pt and Ag wires, respectively, protected by a glass mantle. Ferrocene (Fc) served as the internal reference, added just before the final potential scan. Occasionally, decame-thylferrocene (Fc*) served for this purpose to avoid an overlap of the standard system with the nearby Mo(II)‒NCS oxidation; E1/2 = 0.48 V vs Fc/Fc+ (in THF). The CV samples contained 1 mM analyte and 0.1 M tetrabutylammonium hexafluorophosphate (TBAH) as the supporting electrolyte.

IR and UV-vis Spectroelectrochemistry. Infrared spectroelectro-chemical experiments were performed on a Bruker 70v FT-IR spectrometer, equipped with either a DTLaGS detector (measure-

Page 6: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

ments at 298 K) or linked to an external Bio-rad FTS 60 MCT detec-tor (measurements at 223 K). UV-Vis spectroelectrochemical ex-periments were conducted on a Scinco S-3100 diode array spectro-photometer. The electrochemical response in the form of a thin-layer cyclic voltammogram (TL-CV) at v = 2 mV s-1 was recorded during both IR and UV-vis monitoring of the controlled-potential electrolysis with an EmStat3 potentiostat (PalmSens). The spectro-electrochemical measurements were conducted with an optically transparent thin-layer electrochemical (OTTLE) cell (Spectroelec-trochemistry Reading) in the room-temperature51 and the cry-ostated low-temperature52 versions. Both OTTLE cells were equipped with a Pt minigrid working electrode, a Pt minigrid coun-ter electrode, an Ag wire pseudo-reference electrode and CaF2 win-dows. The room-T OTTLE cell was also used with an Au minigrid working electrode. The SEC samples contained 3 mM analyte and 0.3 M TBAH. Computational studies. Density Functional Theory calculations53 were performed, using the Amsterdam Density Functional pro-gram (ADF).54–56 Geometries were optimized with gradient correc-tion, without symmetry constraints, using the Local Density Ap-proximation of the correlation energy (Vosko-Wilk-Nusair)57 and the Generalized Gradient Approximation (Becke’s58 exchange and Perdew’s59,60 correlation functionals). Unrestricted calculations were carried out for open-shell complexes. Solvent was considered in all geometry optimizations and single-point calculations in ac-cordance with the COSMO approach were implemented in ADF. Tetrahydrofuran was always chosen, except for the PrCN-coordi-nated reduced complexes (Scheme 1) that were modelled with ac-etonitrile. Relativistic effects were treated with the ZORA approxi-mation.61 Triple- Slater-type orbitals (STO) were used to describe all the valence electrons of H, O, C, N, S, and Mo. A set of two polar-ization functions was added to H (single- 2s, 2p), O, C, N, S (single-, 3d, 4f) and Mo (5d, 4f). Frequency calculations were carried out to obtain the vibrational spectra and to check that intermediates were minima in the potential energy surface. Time-Dependent DFT calculations62–65 with spin orbit coupling (SOPERT) were per-formed to determine the excitation energies.66 The energy decom-position analysis (EDA)67,68 was performed on single-point calcula-tions in the gas phase over the solvent-optimized geometries.

The X-ray structures described above were used for the calcula-

tions on parent x,x′-dmbipy complexes 1-3 and all the others were modelled after them. The more important isomers of each species were considered, as well as different spin states when relevant. To check the effect of two possible isomers in solution, frequencies and absorption spectra were calculated for both, and the differ-ences were much smaller than differences to other complexes of the same ligand. Three-dimensional representations of the struc-tures and molecular orbitals were obtained with Chemcraft.69

RESULTS AND DISCUSSION

Characterization and Crystal Structure Analysis. Infra-red spectra of dicarbonyl complexes 1-3 show the two ν(CO) absorption bands close to 1950 and 1870 cm-1 (Table 2). Com-pared with the ν(CO) modes of the unsubstituted-bipy refer-ence, [Mo(η3-allyl)(CO)2(bipy)(NCS)], the wavenumbers of 1-3 do not deviate significantly, reflecting the limited electronic effect of the methyl substitution on the CO π-back-donation. The two CO-stretching wavenumbers are almost identical for 2 and only slightly smaller for 1. The largest ν(CO) drop in the series (up to 5 cm-1) is caused by the inductive effect of the do-nor methyl substituents at 6,6′-dmbipy. The ν(CN) mode of the axial thiocyanato ligand in 1 and 2 also remains positioned near 2080 cm-1 (in PrCN) and 2075 cm-1 (in THF) compared to

the reference bipy complex. The CO π-back-donation in-creases, and thiocyanate π-donation decreases at lower tem-peratures, both in PrCN and THF.

The crystal structures of 1-3 (shown in Figure 1 for 1 and

3, and Figure C-S1 (Supporting Information) for 2) reveal that these complexes all adopt the type A pseudo-octahedral struc-ture encountered in most [M(η3-allyl)(CO)2(L∩L)X complexes] (L∩L = chelating bidentate ligand, X = monodentate anionic lig-and), including [Mo(η3-allyl)(CO)2(bipy)(NCS)],70 also known as the equatorial isomer. The type A structure is sym-metric, characterized by both L∩L donor atoms appearing in the trans position to the carbonyl ligands, defining the equato-rial plane. The less symmetric type B structure (axial isomer), where the η3-allyl and one CO ligand are trans to the L∩L donor atoms, is only seen in rare instances71–74, but is particularly common for larger, more sterically-hindered donor ligands or those with strong π-accepting properties such as the 2,6-xylyl-BIAN (= bis(2,6-dimethylphenyl)-acenaphthenequi-nonediimine) ligand.45,74 As is found in practically all the com-plexes in this family, the open face of the allyl ligand in 1-3 lies over the equatorial CO ligands. While almost linear in the ref-erence bipy complex69, the structure of the NCS‒ ligand is bent in the dmbipy complexes, toward the 4,4ʹ-dmbipy ligand in 1, while away from the 6,6ʹ-dmbipy ligand in 3, forced by packing interactions (Figure 1). The bond angle between the equatorial CO ligands is compressed in 3 (76°) relative to 1 (82°). In the reference bipy complex, this angle is 78°, the mid-way point between the two extremes.69 In addition to this, the two rings in chelated 6,6ʹ-dmbipy in 3 are not co-planar, but have a dihe-dral angle around the NCCN moiety of -7.2°. In 1, the corre-sponding angle is 0°. These observations indicate that the me-thyl groups in the 6,6′-positions exert a stronger steric impact over the rest of the complex. Corresponding crystallographic data for 1-3 are presented in Tables C-S1 and C-S2 (Supporting Information).

Page 7: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

Figure 1. An ORTEP view (50% thermal probability) of

molecular structures of [Mo(η3-allyl)(CO)2)(4,4′-dmbipy)(NCS)] (1, top) and [Mo(η3-allyl)(CO)2)(6,6′-

dmbipy)(NCS)] (3, bottom) determined by single-crystal X-ray analysis. Hydrogen atoms have been omitted for clarity.

Table 1. Redox potentials (V vs Fc/Fc+) of complexes 1 and 3 and their reduction products (see Scheme 1) from cyclic voltammetry at an Au microdisc electrode at 298 K. Table E-S1 (Supporting Information) presents the complete list, including the data for complex 2.

Complex Solvent Mo(II/III) E1/2

R1 Ep,c

R2 Ep,c R2′ Ep,c

O1ʹ Ep,a

R(D) Ep,c

O(D) Ep,a

[Mo(η3-allyl)(CO)2(bipy)(NCS)]c THF 0.20 -1.99 - -2.82 -1.74 -2.52 e

PrCN 0.22 -1.95 -2.59 -2.77 -1.74 e -0.80

PrCN b d -1.88a -2.58b f -1.58 e -0.63

[Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(NCS)] (1)

THF 0.18 -2.06 - -2.78 -1.82 -2.46 -1.18

THF b 0.22 -1.96a -2.53 -2.80a -1.68 e -1.17

PrCN 0.22 -1.99 - f -1.74 -2.36 e

PrCN b 0.22 -1.96a -2.55 f -1.54 e -1.16

[Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)(NCS)] (3)

THF 0.26 -2.02a -2.57 -2.94 -1.84 - -

THF b 0.28 -1.98a -2.60 -2.82a -1.66 - -

PrCN 0.32 -1.93a -2.45 f -1.73 - -

PrCN b 0.28 -1.94a -2.56 f -1.54 - -

a E1/2 value (anodic counter wave observed). b Measured at 195 K. c The reference complex, measured at a Pt electrode.45 d Not measured. e Not observable. f Beyond the accessible cathodic potential window of PrCN/TBAH.

Page 8: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

Scheme 1 General cathodic pathways of complexes 1-3 (= X) based on the evidence from cyclic voltammetry (colored molecular structures). The evidence for radical [X-PrCN] and anion [X-PrCN]‒ (X = 1 and 2) comes solely from IR spectroelectrochemistry.

Color scheme: red - parent complex; pink– corresponding one-electron-reduced radical anion; blue – two-electron-reduced five-coordinate anion; green – one-electron-reduced dimer (tentative structure); purple – three-electron-reduced five-coordinate dian-ion.

DFT calculations53 using the ADF program54–56 were per-

formed on 1-3 in order to understand the role of the methyl substituents and to calibrate the method to identify the species detected in the spectro-electrochemical experiments. The equatorial isomer is indeed always preferred, the energy dif-ference increasing as the methyl substituents approach the metal - 0.67, 0.86, and 5.45 kcal mol-1 for 1, 2, and 3, respec-tively (Tables DFT-S1 and DFT-S2, Supporting Information). Substituents in adjacent positions to the donor N-atoms give rise to a distortion of the coordinated dmbipy ligand in the equatorial isomer (Figure 1) but lead to severe repulsion with the allyl hydrogen atoms. This is reflected in the binding en-ergy of each dmbipy to the {Mo(η3-allyl)(CO)2(NCS)} fragment, which is ~41 kcal mol-1 for 1 and 2, but only 33 and 27 kcal mol-1 for the equatorial and axial isomers of 3. The structural parameters are well reproduced (Tables DFT-S3 and C-S2, Supporting Information).

A fragment decomposition analysis helps to understand

the origin of the lower binding energy of the 6,6′-dmbipy ligand

compared to the 4,4’- and 5,5’-derivatives (Table DFT-S4, Sup-porting Information). Indeed, the ca. 8 kcal mol-1 difference re-sults from two sources. The first one is the extra energy needed to distort the fragments from their optimized geometry to the geometry they have in the complex. The Mo fragment requires ca. 0.5 kcal mol-1, and the dmbipy ligands ca. 2.5 kcal mol-1). The second contribution is the loss of covalent interaction be-tween the two fragments (ca. 5 kcal mol-1). These intrinsic dif-ferences will play a role in the reactivity of the complexes.

Cyclic Voltammetry. Cyclic voltammetry of 1-3 was con-

ducted at room (298 K) and low (195 K) temperatures at an Au microdisc electrode in argon-saturated THF/TBAH (Figures 2-4) and PrCN/TBAH (Figures E-S1 – E-S3, Supporting Infor-mation). Reference measurements at a Pt microdisc electrode are depicted in Figures E-S4 - E-S6 (THF) and E-S7 – E-S9 (PrCN), Supporting Information. The redox potentials obtained for 1 and 3 with the Au microdisc are summarized in Table 1. For complex 2 these data are available in complete Table E-S1 (Supporting Information). Redox potentials measured in the

Page 9: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

series 1-3 on a Pt microdisc are summarised in Table E-S2 (Supporting Information).

The redox behavior of 1 closely resembles that reported45

for the reference complex [Mo(η3-allyl)(CO)2(bipy)(NCS)]. In the anodic region, 1 undergoes reversible, formally Mo-based 1e‒ oxidation producing the cationic species [MoIII(η3-al-lyl)(CO)2(4,4ʹ-dmbipy)(NCS)]+, even in strongly coordinating PrCN. This result proves that the donor ability of the dmbipy ligand is sufficient to stabilize the higher oxidation state. De-spite this, the influence of the methyl substituents on the actual anodic potential is negligible. On the slow cathodic scan at v = 100 mV s-1, 1 undergoes irreversible dmbipy-based reduction to 1•‒, in weakly coordinating THF at Ep,c = -2.06 V (R1 in Figure 1). In PrCN, this irreversible process occurs less negatively at Ep,c = -1.99 V. Reversing the scan direction just behind the ini-tial cathodic wave in either solvent exposes a nearby irreversi-ble anodic wave at Ep,a = -1.82 V (THF) or Ep,a = -1.92 V (PrCN); this process is designated O1ʹ. From a comparison with both the reference bipy complex and the cathodic cyclic voltammo-gram of [Mn(CO)3(bipy)Br]22, O1ʹ can be assigned to the oxida-tion of the 5-coordinate anion [Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)]‒ ([1-A]‒ in Scheme 1), classifying the initial reduction

at R1 as a formally 2e‒ process. The initial 1e‒ reduction of 1 forms unstable 1•‒ that readily loses the axial monodentate lig-and NCS‒ (see the Spectroelectrochemical section for support) and converts to the corresponding 5-coordinate radical, [Mo(η3-allyl)(CO)2((4,4ʹ-dmbipy)] ([1-R)] readily reducible at R1′ < R1 (ECE) to [1-A]‒. This 5-coordinate anion is detectable by its irreversible oxidation at O1′ but also its 1e‒ reduction to [Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)]2‒ ([1-A]2‒) at R2′ (becoming reversible at 195 K, E1/2 = -2.80 V (THF)), similar to the redox behavior of [Mn(CO)3(bipy)]‒ under argon.22 The wave R2′ was

poorly resolved in PrCN at the edge of the accessible potential win-

dow.

The new cathodic wave R(D) in the room-temperature CV

of 1, observed at Ep,c = -2.46 V (in THF) and -2.36 V (in PrCN), can be attributed to the reduction of the Mo‒Mo bonded dimer

[Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)]2 ([1-D] in Scheme 1) formed on the selected time scale from the unreduced parent complex by its electron-transfer reaction with [1-A]‒ electrogenerated at R1 by the irreversible ECE process. The reduction of [1-D]

appears reversible on the CV time scale of seconds; although, ultimately it should reproduce [1-A]‒. This route has been re-ported for reference [Mn(CO)3(bipy)]2.22 However, [1-D] ex-hibits inherent secondary reactivity, revealed by IR spectroe-lectrochemistry in the following section, preventing us to ob-tain a clear spectroscopic evidence for this process. The oxida-tion of [1-D] is visible on the reverse anodic scan in THF as a weak anodic wave O(D) at Ep,a = -1.17 V, beyond the oxidation of [1-A]‒ at O1′ that triggers the ECEC dimerization process similar to that at R1 via 5-coordinate radical [1-R] (Scheme 1). The latter species probably binds back the liberated NCS‒ lig-and; the concomitant oxidation (O1′ < O1) reproduces parent 1 that reacts with yet non-oxidized [1-A]‒ to form dimer [1-D] detected more positively by its oxidation at O(D). Differently from the cathodic dimerization route, the donor solvent (espe-cially PrCN) may play a role in the intimate mechanism of the anodic dimerization.

Figure 2. Cyclic voltammetry of [Mo(η3-allyl)(CO)2)(4,4′-dmbipy)(NCS)] (1), in THF/TBAH (a) at room temperature and (b) 195 K. Scan Rate: 100 mV s-1. Au microdisc electrode. The arrow indicates the initial scan direction.

At sufficiently low temperatures, parent radical anion [1]•‒

formed at R1 becomes stable, resulting in a reversible R1/O1 couple, E1/2 = 1.93 V (THF, Figure 1) or -1.96 (PrCN, Figures E-S1 and E-S7, Supporting Information). [1-A]‒ is then formed only at the newly observed irreversible cathodic wave R2, Ep,c

= -2.53 V (THF) or -2.55 (PrCN), upon rapid dissociation of NCS‒ from 6-coordinate [1]2‒. The quasi-reversible reduction of [1-A]‒ to 5-coordinate dianion occurs again at R2ʹ, Ep,c = -2.85 V (THF). Compared to room-temperature CV, the reverse oxidation of [Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)]‒ is shifted upon

cooling more positively (>100 mV) from the parent reduction at

R1. The anodic ECEC dimerization process is observable at low

temperatures only in THF (a minor O(D) wave at -1.17 V).

Page 10: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

Figure 3. Cyclic voltammetry of [Mo(η3-allyl)(CO)2)(5,5′-dmbipy)(NCS)] (2) in THF/TBAH at (a) room temperature and (b) 195 K. Scan rate: 100 mV s-1. Au microdisc electrode.

The analysis of the redox behavior of 2 under ambient con-

ditions (Figure 3 (THF) and Figures E-S2 and E-S8, Supporting Information (PrCN)) leads to a description that is largely com-parable to that of 1 (Table 1, Tables ES-1 and E-S2 (Supporting Information), and Scheme 1), with only minor differences in the electrode potential values of each redox process. It is note-worthy that the reference complex [Mo(η3-al-lyl)(CO)2(bipy)(NCS)] with the less donating bipy ligand45 showed a partially reversible reduction to the corresponding radical anion at R1 in PrCN at room temperature already at v = 100 mV s-1 while the radical anions of 1 and 2 are more reac-tive and not observable under the same experimental condi-tion, as revealed by the absent counter wave O1.

The redox behavior of 3 (Figure 4, and Figures ES-3 and E-

S9, Supporting Information) strongly differs from that of 1 and 2. The formally metal-based oxidation becomes quasi-reversi-ble and markedly positively shifted to E1/2 = 0.26 V (THF) or 0.32 V (PrCN); the anodic potential in PrCN is more positive by

100 mV than that of the reference complex with the unsubsti-tuted bipy ligand, whereas complexes 1 and 2 oxidize at slightly less positive potentials (Table 1, and Tables E-S1 and E-S2, Supporting Information). In the cathodic range, the re-duction of parent 3 at R1 to radical anion 3•‒ at E1/2 = -2.03 V (THF) or -1.93 V (PrCN) is reversible at v = 100 mV s-1 already at room temperature (Figure 4). This difference has several im-plications for following cathodic path. Unlike 1 and 2, there is no evidence for the formation (absent O1ʹ) of 5-coordinate an-ion [3-A]‒, until the irreversible wave R2 at Ep,c = -2.57 V (THF) or -2.45 (PrCN) is passed, where the latter is directly produced by NCS‒ dissociation from the transient six-coordinate dianion. On the reverse anodic scan initiated beyond R2, the anodic wave O1ʹ is clearly seen at Ep,a = -1.84 V (THF) or -1.73 V

(PrCN). The absence of [3-A]‒ at R1 prevents also the for-mation of dimer [3-D], which is reflected in the absence of its R(D) and O(D) signatures. The absent anodic wave O(D) indi-cates that the dimerization reaction between [3-A]‒ and parent

3 (reformed at O(D)) is inhibited. The 5-coordinate anion is fur-ther reduced to [3-A]2‒ at R2ʹ, Ep,c = -2.94 V (THF). At low tem-peratures, R2ʹ takes on the reversible character and slightly shifts to E1/2 = -2.82 V (Figure 4(b)).

Page 11: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

Figure 4. Cyclic voltammetry of [Mo(η3-allyl)(CO)2)(6,6′-dmbipy)(NCS)] (3), in THF/TBAH (a) at room temperature and (b) 195 K. Scan Rate: 100 mV s-1. Au microdisc electrode.

On electronic grounds, the addition of methyl groups to bipy in the 4,4ʹ-positions (1) and 5,5ʹ-positions (2) increases electron donation toward the Mo∩bipy metallacycle and the axial Mo‒NCS bond, as evidenced by the more negative reduc-tion potentials, destabilizing the corresponding radical anions compared to the reference complex with 2,2ʹ-bipyridine. The magnitude of the R1 potential shift is comparable to that re-ported for [Mo(CO)4(x,x′-dmbipy)] (x = 4-6) and [Re(CO)3(x,x′-dmbipy)Cl] (x = 3-5) vs their respective bipy counterparts.39,75 The redox behavior following the donor substitution in the 6,6′-position (3) is extremely unusual, resulting in the appar-ently stable radical anion formed upon 1e‒ reduction. In addi-tion, it also stabilizes [3-A]‒ and hinders the follow-up dimeri-zation involving this 2e‒ reduced species and parent 3. A simi-lar inhibiting effect on the dimerization was reported by Ku-biak and co-workers for [Mn(CO)3(6,6′-dimesityl-bipy)]‒ stabi-lized by the large substituents in the vicinity of the Mn∩bipy metallacycle.25

Figure 5. DFT-optimized structures of the parent complex [Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(NCS)], 1 (the equatorial iso-mer (top left) and the axial isomer (top right)), 1e‒ reduced rad-ical anion [1]•‒ (bottom left), 5-coordinate radical [1-R] (the SP isomer (left) and the TBP isomer (right)), and 2e‒ reduced 5-coordinate anion [1-A]‒ (bottom right), with the relevant bond lengths (Å).

Computational Studies. The behavior of 1-3 upon electro-chemical reduction, and the nature of the reduction products

formed, as revealed by the preceding CV studies, were investi-gated in detail with quantum chemical calculations (Amster-dam Density Functional).54–56 Their initial 1e‒ reduction of 1-3 leads to the formation of corresponding radical anions; only 3•‒

has been found sufficiently stable at ambient conditions (Figure 4). The geometry optimization (Table DFT-S2, Supporting Infor-mation) reveals that the axial isomer is preferred for [1]•‒ and [2]•‒ by less than 1 kcal mol-1, while the less strained equatorial form remains significantly favored by [3]•‒ (ca. 6 kcal mol-1). The binding energy of the NCS‒ ligand in each anion is 7.8 ([1]•‒

), 6.8 ([2]•‒), and 10.74 ([3]•‒) kcal mol-1, suggesting that thio-cyanate will be less easily released by [3]•‒, in line with the CV observations. The effect of the reduction on the structure is ob-served in the shortening of the central C1‒C1′ bond of the bi-dentate ligand (from 1.476 to 1.428 Å in 1), reflecting the dom-inant localization of the complex LUMO on x,x′-dmbipy, and in the lengthening of the carbonyl C‒O distances (Figure 5, and Figures DFT-S1 and DFT-S2, Supporting Information).

The HOMO of 1 (Figure 6) is bonding between the metal and the -acceptor carbonyls and allyl, but π-antibonding be-tween Mo and the axial -donor thiocyanate, being mainly a Mo orbital. Hence, the 1e‒ oxidation of 1-3 can formally be MoII → MoIII + e‒. Strong σ-donation from x,x′-dmbipy (x = 4-6) is there-fore needed to stabilize [1]+-[3]+, as observed (see the follow-ing SEC section). The LUMO and LUMO+1 of 1 are largely local-ized on 4,4′-dmbipy, both being bonding between C1 and C1′. Shortening of this distance is therefore expected upon the re-duction.

Figure 6. DFT-calculated frontier orbitals in the parent com-plex [Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(NCS)], 1. The HOMO-LUMO (H-L) gap is 1.574 eV.

Page 12: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

The identification of the diverse carbonyl complexes formed along the cathodic pathways can be most effective by analysis of the vibrational stretching modes of the CO ligands (Table DFT-S5, Supporting Information). The calculated wave-numbers are practically independent of the methyl substitu-tion pattern. The absorption maxima of the symmetric and an-tisymmetric modes for 1, 2 and 3 lie at 1882, 1883, 1881 cm-1 and at 1800, 1801, 1800 cm-1, respectively. Frequency calcula-tions need a scaling factor to reproduce experimental values. These calculated values are ca. 0.97 of the experimental ones, in a very good agreement. Interestingly, the trend in the neutral

and singly reduced species is the same (the largest wavenumbers found for 2). In THF, the experimental wavenumbers for 1 and [1]•‒ are 1949, 1869 cm-1 and 1926, 1815 cm-1, respectively (Table 2, in the following section). Application of the 0.97 factor converts them to 1891, 1813 cm-1 and 1868, 1760 cm-1, respec-tively; these values are very close to those calculated in Table DFT-S5 (Supporting Information).

Electronic absorption of 1-3 in the visible spectral region, having a charge-transfer character, was investigated with TDDFT calculations and the results are presented in Support-ing Information (see Table DFT-S6 and the accompanying text). The calculated frontier molecular orbitals of 1-3 are shown in Figure 6, and Figures DFT-S3 and DFT-S4 (Supporting Infor-mation), respectively.

The concomitant loss of the thiocyanate anion from the ini-tial radical anion will afford the transient 5-coordinate radicals [Mo(η3-allyl)(CO)2(x,xʹ-dmbipy)] (x = 4-6; [X-R] in Scheme 1). Their structure has been derived from both parent equatorial and axial isomers; this approach yielded geometries close to a square plane (SP) or a trigonal bipyramid (TBP), respectively. The latter was the most stable one by a negligible 0.4 kcal mol-

1 for [1-R] and [2-R], whilst the square-planar geometry was more strongly preferred for [3-R] (ca. 5 kcal mol-1). Radical [3-R] was less stable than [1-R] and [2-R] by ca. 10 kcal mol-1. The structure of [1-R] exhibits, for the two geometries calculated, larger carbonyl C‒O and 4,4′-dmbipy inter-ring C‒C bond lengths than calculated for [1]•‒, and smaller Mo‒C(allyl) bond lengths (these are merely trends, the absolute differences are small, see Figure 5). These changes indicate that the loss of the NCS‒ anion leads to stronger delocalization of the additional electron from the dominantly 4,4′-dmbipy radical anion in [1]•‒

over the Mo-bis(carbonyl) unit in [1-R] (Figure 7).

Owing to their role in the reactivity characterizing the ca-thodic path, the frontier orbitals of [1-R], as well as the spin density distribution in the 5-coordinate radical, are important. The orbitals of [1-R] are depicted in Figure DFT-S5 (Supporting Information). The spin density in the radicals is delocalized over the metal and the 4,4′-dmbipy ligand, with practically no contributions from the allyl and the carbonyls, as shown for [1-R] in Figure 7 (left), and for [2-R] and [3-R] in Figures DFT-S6 and DFT-S7 (Supporting Information), respectively.

Figure 7. DFT-calculated spin density in the 5-coordinate rad-ical [Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)], [1-R] (left), and precur-sor 6-coordinate radical anion [1]•‒ (right).

The direct 1e‒ reduction of 5-coordinate radicals [X-R] (X = 1, 2, 3), formed at the irreversible cathodic wave R1 (Figure 1(a)), produces 5-coordinate anions [X-A]‒ that may in princi-ple exist in closed-shell singlet or open-shell triplet states. As expected, the former alternative is more stable in all the three cases, by 16.9 and 19.2 kcal mol-1 for [1-A]‒ and [2-A]‒, but by only 11.8 kcal mol-1 for [3-A]‒, emphasizing again the differ-ences exhibited by the 6,6’-dmbipy derivatives. The addition of the second electron leads to longer Mo‒C(allyl) bonds com-pared not only to precursors [X-R] but also to neutral parent compounds X, suggesting a possible allyl loss opening a decom-position pathway. The C‒O bonds lengthen, while the C‒C inter-ring bond in reduced x,x′-dmbipy shortens slightly. The struc-ture of [1-A]‒ is shown in Figure 5; the others for X = 2 and 3 can be found in Figures DFT-S1 and DFT-S2 (Supporting Infor-mation), respectively.

Figure 8. DFT-calculated frontier orbitals of the five-coor-dinate anion [Mo(η3-allyl)(CO)2(6,6′-dmbipy)]‒, [3-A]‒.

The calculated frequencies of parent X shift by 26-30 cm-1 for ν(CO)s and 37-40 cm-1 for ν(CO)a when reduction occurs and

Page 13: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

radical anions [X]•‒ are formed (Table DFT-S5, Supporting In-formation). Interestingly, the loss of the thiocyanate ligand has practically no effect on the stretching frequencies when going from [3]•‒ to [3-R] (less than 5 cm-1 for both modes). This effect on the π-back-donation to the carbonyl ligand is understandable, as the dissociation of the π-donor thiocyanate is compensated by the increased π-delocalization of the added electron from 4,4′-

dmbipy to the (CO)2Mo-dmbipy metallacycle. For the 4,4-dmbipy and 5,5’-dmbipy ligands, the consequence of the thiocyanate dissociation is not so small, since the ν(CO)s mode shifts by 7 and 5 cm-1, respectively (for the axial geometry). As can be seen in Table DFT-S5 (Supporting Information), the reduction from the 5-coordinate radicals to the corresponding anions has more dramatic effects, the shifts reaching ~90 cm-1 for the de-rivatives of 1 and 2, and ~50 for cm-1 those of 3. This difference testifies that the added second electron largely resides on the Mo dicarbonyl unit due to the extensive π-delocalisation of the frontier orbitals in [X-A]‒. This characteristic bonding property is clearly evidenced by the nature of the calculated frontier or-bitals depicted in Figure 8 for [3-A]‒, and Figures DFT-S8 and

DFT-S9 (Supporting Information) for [1-A]‒ and [2-A]‒, respec-

tively. The strongly π-delocalized situation applies in all three cases

for the HOMO and LUMO. The LUMO+1 and LUMO+2 are dom-

inantly x,x′-dmbipy-localized while the HOMO-1 and HOMO-2

reside on the Mo‒dicarbonyl unit. The molecular orbitals of the

five-coordinate anions are indeed very similar in nature to those of

the radical precursors (α orbitals), as can be seen for [1-R] and [1-

A]‒ in Figures DFT-S5 and DFT-S8 (Supporting Information), re-

spectively.

Figure 9. DFT-optimized structures of the 6-coordinate anion [Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(PrCN)]‒, [1-PrCN]‒ (the sta-ble equatorial isomer (right)), and the staggered Mo‒Mo dimer, [1-D] (left), with relevant bond lengths (Å) .

Two other reduced species are also formed, depending on the x,x′-dmbipy ligand. First, the existence of the 6-coordinate anion [Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(PrCN)]‒, [1-PrCN]‒, is not obvious from the preceding CV studies but a strong evi-dence for its participation comes from IR spectroelectrochem-istry described in the following section. This low-temperature complex of PrCN and 2e‒ reduced [X-A]‒ in the equatorial confor-

mation was only observed for [1-PrCN]‒ (Figure 9) and [2-PrCN]‒ (Figure DFT-S10, Supporting Information); the absence of [3-PrCN]‒ (Figure DFT-S10, Supporting Information) may be explained by stronger stabilizing π-delocalization in the closer-to-square pyramidal structure of [3-A]‒ as well as by the steric hindrance by the adjacent methyl substituents. For [1-PrCN]‒, the axial and equatorial isomers, both as diamagnetic or para-magnetic species, were explored. but only the diamagnetic equatorial isomers could be calculated with PrCN coordinated in the original position of the thiocyanate ligand in parent 1 (Figure 9). The 4,4′-dmbipy inter-ring C‒C distance of 1.399 Å

is significantly shorter compared to the values calculated for both diamagnetic [1-A]‒ (1.432 Å) and paramagnetic [1]•‒ (1.426 Å) where 4,4′-dmbipy is considered to be formally 1e‒ reduced. This may lead to the conclusion that both added electrons in [1-PrCN]‒reside largely in the lowest π*(dmbipy) orbital. On the other hand, the calculated wavenumbers of the ν(CO) and PrCN ν(CN) modes in [X-PrCN]‒ are still too low and too high, respec-tively, compared to the experimental values for stable 6-coor-dinate [1-PrCN]‒ (Table 2), which indicates only rather weak interactions of [X-A]‒ with PrCN, reflected also in positive bind-ing energies with regard to the isolated components.

Five-coordinate radical and anionic species, [X-R] and [X-

A]‒, in the TBP geometry do not have obvious occupied frontier orbitals to form a Mo‒Mo bond, but the SP conformers feature the SOMO and HOMO, respectively, with a strong contribution of 4dz2, as reflected in the orbital contour plots (Figure DFT-S11, Supporting Information, for X = 1). Therefore, it is easy to imagine two radicals forming a Mo‒Mo bond, in [1-D], by shar-ing electrons. This is equivalent to the formation of the dimer by the reaction between the 2e‒ reduced anion and the neutral parent complex after the loss of the thiocyanate (via donation of 2e‒ from [1-A]‒ to the empty orbital of 1). Dimer [1-D] ob-tained (Figure 9), displays a very long Mo‒Mo bond but is a true intermediate (see the ν(CO) wavenumbers in Table DFT-S5, Supporting Information). Other bond lengths are within nor-mal values. A similar dimer could be obtained for 2 (Figure DFT-S10, Supporting Information). A definitive answer to the true nature of the dimer would require a more extensive anal-ysis of possible alternatives and more experimental infor-mation.

The 2e‒ reduction of 1 and 3 to corresponding 5-coordinate anions [X-A]‒ (Scheme 1) was also carried out under a CO2 at-mosphere, and there was an evidence of a different reactivity (see the following SEC section). Therefore, we used DFT calcu-lations to look for derivatives of the 5-coordinated anion with CO2, both as equatorial and axial isomers of the adduct [1-CO2]‒

. The equatorial isomer (Figure 10) was more stable by 7.2 kcal mol-1 than the axial one (Table DFT-S1, Supporting Information).

Figure 10. DFT-optimized structure of [Mo(η3-al-

lyl)(CO)2(4,4ʹ-dmbipy)(CO2)]‒, [1-CO2]‒. The narrow O‒C‒O angle in [1-CO2]‒ of ca. 125 is con-

sistent with a strong electronic reduction accompanying the coordination of CO2 caused by the electron transfer from for-mally dianionic 4,4′-dmbipy seen, for example in [1-PrCN]‒.

This is evidenced by a strong elongation of the C1‒C1′ inter-ring distance from 1.399 Å in the latter anion to 1.466 Å in [1-CO2]‒. The latter value is very close to 1.476 Å for neutral 4,4′-dmbipy in 1. The C=O bonds in the reduced CO2 ligand are elongated to 1.273/1.275 Å, which reflects their weakening. There is still significant electron density on the two carbonyl ligands, as re-vealed by the calculated CO-stretching wavenumbers of 1831

Page 14: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

and 1714 cm-1 for the symmetric and asymmetric mode, re-spectively.

Infra-red and UV-vis absorption spectroelectrochemis-

try at low temperature. Molecular spectroelectrochemistry is commonly used in combination with conventional cyclic volt-ammetry (CV) to assign major redox products. In this work IR and UV-vis spectroscopy, in combination with thin-layer cyclic voltammetry (TL-CV), is very convenient to unravel the reduc-tion paths of complexes 1-3 in association with the insight from the conventional cyclic voltammetry and the predictive power of the relevant DFT data in the preceding sections. The focus is placed on analysis of the IR stretching pattern and wavenumbers of the carbonyl ligands, the CN-stretching modes of the NCS‒ and PrCN ligands, and the electronic absorp-tion of the reduced x,x′-dmbipy ligands.

All three parent complexes are smoothly oxidized at 293 K

to the corresponding stable (radical) cations. The correspond-ing CO- and CN-stretching wavenumbers and visible electronic absorption are summarized in Table E-S3 (Supporting Infor-mation). The IR spectral changes are supportive of the depop-ulation of the π*(Mo‒NCS) HOMO of the parent complexes, with a strong impact on the Mo-to-CO π-back-donation, strongly reduced in the formally Mo(III) products. It is im-portant to note that the thermal stability of [X]+ is significantly higher compared to the cationic complex with the unsubsti-tuted bipy ligand45, proving the stronger donor ability of all three dimethylated x,x′-dmbipy ligand isomers. This result in-dicates that any major differences in the cathodic behavior of 1-3 should have their origin in the steric properties of the metallacycle imposed by the methyl substituents at x,x′-dmbipy (x = 4-6).

It is prudent to start the description of the cathodic behav-

ior in the studied series of complexes with the low-tempera-ture spectroelectrochemistry of 3 that displays arguably the most straightforward redox behavior, as revealed by cyclic voltammetry (Figure 4). This is because [Mo(η3-al-lyl)(CO)2(6,6ʹ-dmbipy)(NCS)]•‒ (3•‒) is perfectly stable in PrCN at low temperature; then, passing the reversible reduction step R1 at 223 K, only the radical anion is detected. The sole product is easily assignable from the distinctive high-energy shift of the ν(CN) band of NCS‒ from 2069 to 2089 cm-1 and the accompa-nying the low-energy shift of the two ν(CO) bands from 1943 and 1859 cm-1 to 1919 and 1829 cm-1, respectively (Figure 11(a)). These values show reasonable agreement with those procured from the DFT calculations (Table DFT-S5 (Support-ing Information) and Table 2, scaling factor 0.97). The corre-sponding UV-vis-NIR spectral changes recorded during the in-itial reduction (Figure E-S10, Supporting Information) confirm the presence of the 6,6′-dmbipy radical anion in 3•‒ by its char-acteristic intra-ligand ππ* absorption at 396 nm, bifurcated π*π* at 489 and 518 nm, and low-lying π*π* >600 nm37,76, in line with the bipy-based LUMO of parent 3 (Figure DFT-S4, Supporting Information).

Radical anion 3•‒ was smoothly reduced in the following,

irreversible 1e‒ cathodic step at R2 generating the 5-coordi-nate anion [Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)]‒, [3-A]‒ (Scheme 1). This process is recognized by the appearance of free NCS‒ absorbing at 2050 cm-1, and two low-lying broad ν(CO) bands at 1797 and 1700 cm-1, which are characteristic of a negatively charged dicarbonyl 5-coordinate species. The experimental

wavenumbers are in a very good agreement with the DFT val-ues calculated for the equatorial square-pyramidal structure (Table 2) stabilized by strongly delocalized π-bonding in the metallacycle. It is the first time that reliable IR ν(CO) values are reported for 2e‒ reduced [Mo(η3-allyl)(CO)2(α-diimine)]‒. In the published45 pioneering work, the IR absorption of the more reactive anion [Mo(η3-allyl)(CO)2(bipy)]‒ (detected by conven-

tional cyclic voltammetry) was confused with products of thermal

dimerization reactions (Scheme 1, and spectroelectrochemistry be-

low). In the visible spectral region, [3-A]‒ exhibits the charac-teristic absorption between 450-600 nm (Figure E-S11, Sup-porting Information) belonging to ππ* electronic excitation77 between the delocalized frontier orbitals of the Mo‒dmbipy metallacycle (Figure 8).

In the preceding CV section, radical anions 1•‒ and 2•‒ were

found to convert at ambient conditions within seconds to the corresponding five-coordinate anions already at the initial ca-thodic wave R1 (Figures 2(a) and 3(a), respectively). For 3•‒,

however, this ECE process is too slow to be detected on the CV

time scale (Figure 4(a)). Therefore, the reduction of 3 at R1 and subsequent reactivity of its radical anion were studied by IR SEC at variable temperatures. At 223 K, 3•‒ remained stable.

However, gradual temperature elevation to 273 K at the constantly

applied cathodic potential R1 resulted in its complete conversion

to 2e‒ reduced [3-A]‒ (Figure 12). This is the high-T ECE equiv-

alent of the low-T EEC path at the cathodic potential R2 (Figure

11(b)). Reasons for the unusually high stability of 3•‒ are briefly

discussed in the preceding DFT section.

Page 15: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

Figure 11. IR SEC monitoring of (a) the initial 1e‒ reduction of [Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)(NCS)], 3 (↓) at R1 to stable 3•‒ (↑), and (b) the following reduction at R2 showing the con-version of 3•‒ to [3-A]‒ (↑) and free NCS‒. Conditions: a cry-ostated OTTLE cell, PrCN/TBAH, 223 K.

Compared to 3, the cathodic behavior of 1 under the same

SEC conditions is affected by both the higher thermal reactivity of 1•‒ and the different steric properties of the 4,4′-dmbipy ligand. Despite appearing stable on the CV timescale (Figure 2(b)), the low temperature of 223 K set in the OTTLE cell was insufficient to completely stabilize the radical anion. IR monitoring of the electrolysis at R1 revealed simultaneous appearance of 1•‒ as the primary reduction product, and another dicarbonyl com-plex assigned as the PrCN-substituted 6-coordinate anion, [1-

PrCN]‒ (Figure 13, Scheme 1) that replaces in the cooled

strongly coordinating solvent the 2e‒ reduced 5-coordinate an-

ion, [1-A]‒, in contrast to 3 and [3-A]‒. Radical anion 1•‒ is once again identified by the ca. 5 cm-1 high-energy shift of the ν(CN) mode of coordinated NCS‒ and the low-energy shift of the two ν(CO) modes by some 20-30 cm-1, in this case from 1946 and 1862 cm-1 to 1922 and 1832 cm-1. The presence of the singly reduced 4,4ʹ-dmbipy ligand in 1•‒ has been confirmed by the par-

allel UV-vis monitoring of the reduction course (Figure E-S12,

Supporting Information).

Table 2. IR and UV-vis absorption data for the complexes [Mo(η3-allyl)(CO)2(x,x′-dmbipy)(NCS)], 1 (x = 4), and 3 (x = 6), and their reduction products (Scheme 1). See complete Table E-S4 (Supporting Information) including complex 2 (x = 5) and reference [Mo(η3-allyl)(CO)2(bipy)(NCS)].

Complex ν(CO)/ cm-1 Calculatedh ν(CO)/ cm-1

ν(CN)/ cm-1 Calculatedh ν(CN)/ cm-1

λmax/ nm

[Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(NCS)] a 1949, 1866 - 2080 - 470

[Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(NCS)] a,c 1946, 1863 - 2084 - -

[Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(NCS)] b 1949, 1869 1882, 1800 2076 2056 -

[Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(NCS)] b,d 1948, 1868 - 2078 - -

[Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)(NCS)] a 1946, 1862 - 2078 - 470

[Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)(NCS)] a,c 1944, 1860 - 2082 - -

[Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)(NCS)] b 1948, 1866 1881, 1800 2074 2054 -

[Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(NCS)]•‒

a,c

1922, 1832 1855, 1763 2089 2070 388, 494, 538, > 600

[Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(NCS)]•‒ b,d 1926, 1831 2084

[Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)(NCS)]•‒

a,c

1920, 1829 1855, 1764 2089 2069 396, 489, 518, > 600

[Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)(NCS)]•‒ b 1925, 1834 - 2089 - -

[Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)(NCS)]• a 1923, 1833 - - -

[Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(PrCN)]‒

a,c

1896, 1797 1797, 1705 2148i 2229 350

[Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)(PrCN)]‒ a 1904, 1789 1816, 1715 2148i 2247 -

[Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)]2 b 1891,1776,

1759

1775,1787,

1844,1858

- - -

[Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)]‒ b,d 1815, 1722 1741, 1656e -

[Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)]‒ a,c 1797, 1700g 1803, 1702f - - 420, 550

[Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)]‒ a 1807, 1700 - - - -

[Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)]‒ b 1792, 1680j - - - -

a Measured in PrCN, b Measured in THF. c Measured at 223 K. d Measured at 268 K. e Calculated axial isomer. f Calculated equatorial isomer. g Broad absorption bands. h Some DFT data from Table DFT-S5 (Supporting Information) are reproduced here for clarity. i CN stretching mode of the PrCN ligand. j Values measured with an Au minigrid working electrode; the ν(CO) values for a Pt minigrid are 1801 and 1717 cm-1.

Page 16: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

Figure 12. Controlled-potential IR SEC monitoring of the ther-

mal conversion of [Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)(NCS)]•-, 3•‒

(↓) to [3-A]‒ (↑) and free NCS‒ between 223-293 K. Conditions: PrCN/TBAH, a cryostated OTTLE cell.

The new axial PrCN ligand in the second observed reduction

product, [1-PrCN]‒, is typically recognized in the IR spectrum by its low-intensity ν(CN) absorption at 2147 cm-1 (Figure 13). The liberated NCS‒ ligand absorbs again at 2050 cm-1. The accompa-nying ν(CO) bands at 1898 and 1797 cm-1 of [1-PrCN]‒ are much narrower compared to those of 5-coordinate [3-A]‒ (Figure 11(b)) and [1-A]‒ formed in THF at 268 K (see below, Figure 14). A significant difference in ν(CO) wavenumbers and band width was also encountered for related 2e‒ reduced anions [Re(CO)3(bipy)]‒ and [Re(CO)3(bipy)(PrCN)]‒.78 The ν(CO) wavenumbers do not correspond with any 1e‒ reduced radical species, being much smaller than those of 1•‒ (Table 2). On the

other hand, they are much larger compared to [1-A]‒ (Table 2) be-

cause the second added electron in the latter π-delocalized anion

largely resides on the Mo(CO)2 moiety while in 6-coordinate [1-

PrCN]‒ on the (formally dianionic) bipy ligand (see the preced-ing DFT section). The UV-vis spectra corresponding to the IR SEC situation in Figure 13 also indicate the presence of two re-duced species, one of which must be 1•‒ absorbing at 388, 494, 538 and >650 nm. The electronic absorption arising around 350 nm is then assigned to [1-PrCN]‒; importantly, the character-

istic metallacycle ππ* absorption of [X-A]‒ at 500-600 nm (Table

2) is absent in this case (Figure E-S12, Supporting Information), in

line with the IR SEC results. The comparison of the reduction paths

of complexes 1 and 3 clearly reveals that the 6,6′-dmbipy ligands

hinders coordination of PrCN to the Mo center of [X-A]‒, probably

not only for steric but also electronic reasons. Data in Table DFT-

S5 (Supporting Information) testify that [1-A]‒ prefers a different

structure than [3-A]‒, with an impact on the π-delocalization in the

Mo‒dmbipy metallacycle stabilizing the 5-coordinate geometry of

the 2e‒ reduced species.79

Conducting the reduction of 1 in weakly coordinating THF

at 268 K induces a very similar behavior as described above for 3 in the chilled PrCN electrolyte. In addition to intermediate 1•‒

, the main product formed at the cathodic wave R1 is now the 5-

coordinate anion [1-A]‒ absorbing in the ν(CO) region at 1815 and 1722 cm-1. The wavenumbers are much smaller compared to 6-coordinate [1-PrCN]‒ (see above). On the other hand, both values

are slightly larger than those determined for 5-coordinate [1-A]‒ in

chilled PrCN and in THF at ambient temperature (Table 2), while

they should be smaller according the DFT calculations. This differ-

ence can be explained by the formation of a weak adduct of [1-A]‒

with THF, causing slightly decreased electron density on the car-

bonyl ligands. Much larger ν(CO) wavenumbers can be anticipated

in this regard for ordinary 6-coordinate anion with coordinated

THF, [1-THF]‒.80

Figure 13. IR SEC monitoring of the electrochemical reduction of [Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(NCS)], 1 (↓) at the cathodic wave R1, forming 1•‒ (↑) and 6-coordinate [1-PrCN]‒ (*). Condi-tions: PrCN/TBAH, 223 K, cryostated OTTLE cell.

The course of the reduction of 2 at R1 in PrCN at 223 K (Fig-

ure E-S13, Supporting Information) greatly combines features observed for 1 and 3, with the parallel production of radical an-ion 2•‒ (ν(CO) at 1921, 1831 cm-1), and 6-coordinate anion [2-PrCN]‒ (ν(CO) at 1897, 1789 cm-1) in a thermal equilibrium with five-coordinate anion [2-A]‒ (ν(CO) 1789, 1692 cm-1). This product representation was confirmed by the parallel UV-vis monitoring (Figure E-S14, Supporting Information). The visible region is dominated by the characteristically broad absorption of the 5-coordinate anion at around 600 nm. This experiment presents the most convincing evidence for the transition posi-tion of the 5,5′-dimethyl substitution at the redox active bipy ligands in the studied series.

Figure 14. IR SEC monitoring of the reduction of [Mo(η3-al-lyl)(CO)2(4,4ʹ-dmbipy)(NCS)], 1 (↓) at the cathodic wave R1,

Page 17: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

generating 1•‒ (↑) and five-coordinate [1-A]‒ (*) as a weak ad-duct with THF. Conditions: THF/TBAH, 268 K, cryostated OTTLE cell.

Infra-red spectroelectrochemistry at ambient tempera-

ture. The discussion of the generally more complicated ca-thodic paths and reactivity induced by the added electrons at room-temperature will be based on IR spectroelectrochemical results obtained in weakly coordinating THF and strongly coor-dinating PrCN. It focuses on participation of peculiar dimeriza-tion reactions of 1 and 2, which are inhibited by cooling the electrolyte solution (see the preceding part).

It is convenient to start the description and discussion of the

cathodic behavior at 293 K again with 3 bearing the sterically demanding 6,6′-dmbipy ligand. Notably, the reduction path in THF mirrors the temperature-controlled conversion of 1e‒ re-duced 3•‒ in PrCN at R1 to 2e‒ reduced 5-coordinate [3-A]‒ via [3-R] (Figure 12). Thus, during the reduction of 3 at R1, only a small amount of 3•‒ (ν(CN): 2087 cm-1, and ν(CO): 1924, 1835 cm-1) is initially detected (Figure 15), the transient species be-ing readily converted to stable 2e‒ reduced [3-A]‒ (ν(CO): 1792, 1680 cm-1). This process may involve an intermediate (a contact

species) absorbing at 1900 cm-1, differently from the instant low-

temperature ECE path. It is worth mentioning that the wave-numbers of [3-A]‒ in this experiment are somewhat smaller than those measured in PrCN at 223 K (Table 2). This observa-tion indicates a weak interaction between [3-A]‒ and PrCN at the

low temperature. The spectral variability in the ν(CO) region is thus

a common feature in the whole series of [X-A]‒.

Figure 15. IR SEC monitoring of the initial reduction of [Mo(η3-

allyl)(CO)2(6,6ʹ-dmbipy)(NCS)], 3 (↓) at R1 to 5-coordinate [3-A]‒ (↑) via intermediate 3•‒ (↑↓). Conditions: THF/TBAH, 293 K, OTTLE cell.

The cathodic path of 3 leading to [3-A]‒ in PrCN at 293 K

(Figure E-S15, Supporting Information) is less straightforward than in THF. The initially detectable reduced species shows two ν(CO) bands at 1923 and 1797 cm-1, being accompanied by free thiocyanate absorbing at 1950 cm-1. The absence of the ν(CN) absorption of the coordinated thiocyanate (typically above the parent absorption at 2074 cm-1) excludes the formation of 3•‒. The higher stability of the reduced dicarbonyl complex formed,

compared to 3•‒ in THF (Figure 15), supports its tentative as-signment as the solvento radical [Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)(PrCN)], [3-PrCN]. As anticipated, the concomitant re-duction of this complex produces [3-PrCN]‒ (ν(CO) at 1904 and 1789 cm-1) that was not detected along the straightforward ECE route at 223 K (Figure 12). It is therefore not surprizing that the 2e‒ reduced 6-coordinate anion is unstable also at 293 K and converts gradually to 5-coordinate [3-A]‒ (ν(CO) at 1807 and 1700 cm-1). The described peculiar cathodic path also includes a

contact species (ν(CO) absorption at 1832 and 1733 cm-1), aris-

ing from the co-existence of [3-PrCN] and [3-A]‒. This absorbance

disappears during the follow-up reduction of [3-PrCN] to [3-PrCN]‒, the latter converting ultimately to genuine [3-A]‒. The

main factor responsible for the different reduction paths of 3 in

PrCN at 223 and 293K, and in THF at 293 K, is the (in)stability of

3•‒ as the primary 1e‒ reduction product. It is also noteworthy that

the ν(CO) wavenumbers of [3-PrCN] (1923, 1797 cm-1) are very similar to those calculated (using the scaling factor of 0.97) for the equatorial isomer of 5-coordinate radical [3-R] (Table DFT-S5, Supporting Information). This means that differently from THF, where [3-R] is directly reducible to [3-A]‒, the 5-coordinate

radical becomes stabilized by a weak bonding interaction with

PrCN. There is no evidence for its secondary dimerization (Scheme

1), pointing to the protective role of the 6,6′-dmbipy ligand.

Figure 16. IR SEC monitoring of the 1e‒ reduction of

[Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(NCS)], 1 (↓) at R1 to interme-diate 1•‒(↓↑) concomitantly converting to [Mo(η3-al-lyl)(CO)2(4,4ʹ-dmbipy)]2-related dimer [1-D′] (↑). Conditions: THF/TBAH, 293 K, OTTLE cell.

Reducing 1 in THF at 293 K, already the first cathodic step

at the electrode potential of R1 shows a striking difference from the reduction of 3 that converts under the same conditions of thin-layer electrolysis exclusively to 5-coordinate anion [3-A]‒ (typically broad ν(CO) bands at 1792, 1680 cm-1). For 1, the cor-responding anionic species, [1-A]‒, was only detected at 268 K as a weak adduct with THF (Table 2). At room temperature, in-stead, no [1-A]‒ was formed at R1 and the dominant product (apart

from minor 1•‒ still detectable in the thin-solution layer of the

OTTLE cell) showed an unprecedented IR spectral pattern of three narrow ν(CO) bands at 1891, 1778 and 1759 cm-1 (Figure 16). These values of absorption maxima reasonably agree with the ν(CO) wavenumbers obtained with DFT for the staggered structure of the Mo‒Mo bound dimer [Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)]2, [1-D] (Figure 9, and Tables DFT-S5 and E-S4, Sup-porting Information). The calculated intensity pattern of the

Page 18: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

four ν(CO) modes for [1-D], 1858 (medium), 1844 (strong), 1787 (strong), and 1775 (weak), has been encountered81 for the tetracarbonyl dimer [Ru(CO)2(iPr-DAB)(Me)]2 (iPr-DAB = 1,4-diisopropyl-1,4-diaza-buta-1,3-diene)]. For 1, the exact mo-lecular structure of the dimeric end-product showing the differ-ent ν(CO) intensity pattern (1-D'), is unknown at present. How-ever, it is assumed to be closely related to the metal‒metal bound dimers represented by [Ru(CO)2(iPr-DAB)(Me)]2 and [Mn(CO)3(bipy)]2, resulting from the general ECEC cathodic route shown also in Scheme 1, which has been based on the ev-idence gained from faster cyclic voltammetry. A strong argu-ment for a different, reinforced structure of [1-D′] is its subse-quent reduction that does not convert it to 5-coordinate anion, [1-A]‒, which is a typical ECE step for both Ru‒Ru and Mn‒Mn dimers, but most likely to singly reduced [1-D′]‒. For [Mo(η3-allyl)(CO)2(bipy)(NCS)], with a cathodic behavior closest to that of 1, the equivalent of [1-D′] absorbs in the IR CO-stretching re-gion at 1891, 1778 and 1757 cm-1.45 The subsequent reduction shifts this assembly down to 1844, 1723 and 1700(sh) cm-1. The latter wavenumbers were assigned incorrectly to [Mo(η3-al-lyl)(CO)2(bipy)]‒, which is actually hardly seen in THF at 293 K, similar to [1-A]‒. It remains a challenge to prepare, isolate and characterize [1-D′] in a follow-up study.

Reducing 1 at room temperature in PrCN (Figure E-S16,

Supporting Information), dimer [1-D′] is again the dominant product, suggesting that under ambient conditions there is a strong driving force to form this species. The intimate mecha-nism is slightly different from that in THF, with [1-PrCN], and [1-A]‒ in a weak contact with PrCN, being involved as intermediate

species. Much like at low temperature in PrCN (see above), reduc-

tion of 2 in THF at 293 K (Figure E-S17) represents an interme-diate case between the routes of 1 and 3, producing ultimately a mixture of both the 5-coordinate anion and a dimer. The [Mo(η3-allyl)(CO)2(5,5ʹ-dmbipy)]2-related dimer, [2-D′], is re-vealed by the three ν(CO) bands at 1892 (sharp), 1780 and 1761 cm-1; the ν(CO) wavenumbers of [Mo(η3-allyl)(CO)2(5,5ʹ-

dmbipy)]‒ (1820, 1730 cm-1) point to its weak interaction with THF. A very similar cathodic path is followed by 2 in PrCN (Fig-ure E-S18).

Cyclic Voltammetry and IR Spectroelectrochemistry

under a CO2 atmosphere. The CV study of 1 and 3 was re-peated in CO2-saturated weakly coordinating THF at both Pt and Au cathodes, to probe the catalytic activity of the complexes towards CO2 reduction and identify the catalyst. For both com-plexes the initial cathodic wave R1 (Table 1 and Tables E-S1 and E-S2, Supporting Information) remains unchanged, excluding any thermal reaction or adsorption effects. Inspecting the CV of 1 at an Au cathode in CO2-saturated THF (Figure 17), the reduc-tion at R1 does not produce any detectable 5-coordinate anions [1-A]‒ that would be reoxidized (under Ar) at the anodic wave O1′ (Figure 2). The adduct of the anion with CO2, [1-CO2]‒, has been calculated (Figure 10) and analyzed with DFT. Some cata-lytic current enhancement is not seen already at R1 but at a more negative potential close to the reduction of dimer [1-D] at R(D). This observation indicates that either further reduction of [1-CO2]‒ or its transformation product at a potential close to R(D) is needed to trigger the catalytic process, or that the di-merization seen for [1-CO2]‒ in THF under argon cannot be in-hibited under CO2 and the concomitant dimer reduction is

needed to regenerate catalyst [1-CO2]‒. The first route resem-bles tricarbonyl Mn‒iPr-DAB (iPr-DAB = 1,4-diisopropyl-1,4-di-aza-buta-1,3-diene) complexes forming 2e‒ reduced 5-coordi-nate anions via an ECE mechanism, which react with CO2 to form a Mn‒bicarbonate intermediate reducible some 600 mV more negatively than the parent complex, to restore the cata-lytic activity.82 IR monitoring of the electrolysis at R1 was needed to collect a supporting evidence for either pathway, see below.

Figure 17. Cyclic voltammetry of [Mo(η3-allyl)(CO)2(4,4ʹ-dmbipy)(NCS)] (1) in CO2-saturated (red) or Ar-saturated (black) THF/TBAH at an Au microdisc electrode. Scan rate: 100 mV s-1.

Page 19: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

Figure 18. Cyclic voltammetry of [Mo(η3-allyl)(CO)2(6,6ʹ-dmbipy)(NCS)] (3) in CO2-saturated (red) and Ar-saturated (black) THF/TBAH at Au microdisc (a) and Pt microdisc (b) electrodes. Scan rate: 100 mV s-1.

The cathodic behavior of 3 in THF in the presence of excess CO2 on the short time scale of cyclic voltammetry is straightfor-ward, given the relative stability of singly reduced radical ani-ons [3]•‒ formed at R1 and further reducible to [3-A]‒ at R2 (Fig-ure 18). Similar to [1-A]‒, also [3-A]‒ reacts with CO2 to form in-itially [3-CO2]‒, triggering the catalytic CO2 reduction at R2. The higher catalytic current on Au (Figure 18a) compared to Pt (Fig-ure 18b) indicates an involvement of the cathodic surface in the catalytic process. The increased catalytic efficiency may result from a strong stabilizing interaction between the gold surface and [3-A]‒ which has recently been proven for the [Mo(CO)4(bipy)] family of catalysts.42 Compared to [Mo(η3-al-lyl)(CO)2(bipy)(NCS)],43 the anionic catalyst derived from 3 op-erates with higher efficiency – with a higher catalytic current flowing at a lower overpotential. This difference can be as-cribed to a higher stability of [3-A]‒, as [Mo(η3-al-lyl)(CO)2(bipy)]‒ was hardly detectable by in situ IR spectros-copy. In this case, IR spectroelectrochemistry of 3 on the time scale of minutes was expected to provide an evidence for the catalytic activity of [3-A]‒ towards CO2 already at the cathodic wave R1 (in line with Figure 15).

The reduction of 1 at R1 in CO2-saturated THF monitored by

infrared spectroscopy led to a mixture of detectable products.

The ν(CO) absorption at 1893 and 1760 cm-1 almost certainly belongs to dimer [1-D′] (Figure 16), which supports the dimeriza-

tion process seen with cyclic voltammetry (Figure 17). As ex-pected, there is no 5-coordinate [1-A]‒ observed in the IR spectra

at room temperature (Figure 14). However, compared to Figure 16,

representing the cathodic route of 1 in THF under argon, the ν(CO) absorptions at 1910 and 1802 cm-1, seen in Figure E-S19 (Sup-porting Information), are not much different from the wave-numbers predicted for the 6-coordinate anion [1-CO2]‒ (1905, 1780 cm-1), using the difference between the ν(CO) values measured and calculated for [1-A]‒ (Table 2, 90 and 58 cm-1), and the ν(CO) values calculated for [1-CO2]‒. Another reasoning, based on comparison of ν(CO) measured and calculated for [1]•‒ (Table 2) and measured for Some bonding interaction between [1-A]‒ and measured for [1-CO2]‒ (see above) predicts the val-ues calculated for [1-CO2]‒ to be 1839 and 1735 cm-1, which are reasonably close to the values of 1831 and 1714 cm-1 calculated for the DFT-optimized structure of [1-CO2]‒ depicted in Figure 10. The interaction between [1-A]‒ and CO2 has also been re-vealed by conventional cyclic voltammetry (Figure 17). In Fig-ure ES-19 (Supporting Information), there is a concurrent ap-pearance of small bands at 1674 and 1641 cm-1 indicating the formation of free bicarbonate anion (and, indirectly, CO). At the same time, a small amount of free formate (1600 cm-1) starts to appear from the very beginning of the thin-layer electrolysis of 1 at R1. The limited reductive conversion of dissolved CO2, tes-tified by only a small decrease in the intensity of the satellite 13CO2 peak, is in agreement with the relative stability of [1-CO2]‒

under the aprotic conditions. The formation of [1-D′] can also be considered as an inhibiting factor.

In argon-saturated THF at 293 K, 3 forms at R1 2e‒ reduced

[3-A]‒ with the smallest IR ν(CO) wavenumbers of 1792 and 1680 cm-1 indicating the absence of any interaction with the weak donor solvent (Figure 15). The presence of CO2 in the THF electrolyte has some impact on the on the ν(CO) wavenumbers of the five-coordinate anion, which slightly increase to 1801/1792 and 1697/1673 cm-1 (Figure E-S20(a), Supporting Information). The behavior does not correspond with the for-mation of [3-CO2]‒ seen above for 1 with a less hindered Mo‒

dmbipy metallacycle. Rather, [3-CO2]‒ interacts only weakly with

CO2, similar to PrCN at variable temperature (Figures 11 and 12).

Moving the cathodic potential from R1 more negatively, towards

R2 (in the ordinary CV of 3), results in IR spectral changes (Figure

E-S20(b), Supporting Information) revealing some catalytic con-

version of CO2 to free formate and CO (flanked by the characteristic

absorption of the bicarbonate anion). The ν(CO) absorption of [3-CO2]‒ becomes diminished while there is a continuous growth of ν(CO) absorption labelled by asterisk in Figure E-S20(b) (Supporting Information), which most likely corresponds to the inactive dimer, [3-D′], seen also for 1, representing a deactiva-tion route.

CONCLUSIONS

The position of the methyl substituents in the studied series of x,x′-dmbipy (x = 4-6) complexes plays a key role in determin-ing the cathodic paths at variable temperature. There is a direct linear relationship between the position of the methyl group and the stability of both the 1e‒ reduced primary radical anion and 2e‒ reduced 5-coordinate anionic complex increasing in the direction of x = 4 < 5 < 6. This trend reflects an interplay of both steric (dmbipy planarity, Mo‒NCS packing) and electronic (Mo‒

NCS bond strength, delocalized π-bonding) factors. At the ca-

Page 20: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

thodic wave R1, the reduction of 1 produces the unstable radi-cal anions and the five-coordinate anions. The latter product converts to an unreactive dimer that may also exist in its re-duced form; the molecular structure of the end-product, seen in both THF and PrCN, remains to be elucidated. This behavior ex-cludes an unsupported Mo‒Mo bond between 5-coordinate radical units. Radical anions derived from 3 are most stable, surprisingly more than the congeners with unsubstituted bipy as a weaker donor ligand; at ambient temperature they trans-form exclusively to the 5-coordinate anions (in a CE step) that hardly bind donor solvents or dimerize. The key differences in the cathodic paths of 1 and 3, in particular the different coordi-nating ability of 2e‒ reduced [X-A]‒, persist also in the presence of CO2 dissolved in the thin solution layer of the electrolyte. Fur-ther differences and enhanced catalytic activity can be expected upon the addition of Brønsted or Lewis acids, to complete the

comparative study stimulated by the development in the field of the

electrocatalytic CO2 reduction with [Mn(CO)3(R-bipy)X]. Varia-

tion of the α-diimine nature and backbone substitution, as well as probing substituents on the allyl ligand, are a few possible

routes that warrant attention to continue the investigation of the pe-

culiar redox behavior, reactivity and catalytic activity of this chal-

lenging yet largely unexplored family of organometallic com-

pounds.

ASSOCIATED CONTENT

Supporting Information

The following three sections with independent numbering of Tables and Figures. Crystallographic_data (PDF) Supplementary_electrochemical_data (PDF) Supplementary_DFT data (PDF)

AUTHOR INFORMATION

Corresponding Author

E-mail: [email protected]

ORCID

F. Hartl: 0000-0002-7013-5360 M. J. Calhorda: 0000-0002-6872-3569 A. M. Chippindale: 0000-0002-5918-8701

Notes The authors declare no competing financial interest.

ACKNOWLEDGEMENTS

This work was jointly funded by the EPSRC and Spectroelectro-chemistry Reading (a spin-out company of the University of Reading – F.H.). F.L.P.V. thanks Erasmus+ (EU) for her Train-eeship. The University of Reading is acknowledged for the pro-vision of the Chemical Analysis Facility (the CAF lab) – NMR and X-ray diffraction analyses. Mr Nick Spencer (University of Read-ing) is thanked for his help in collecting the single-crystal X-ray diffraction data. M.J.C. thanks the Fundação para a Ciência e a Tecnologia, Portugal, for financial support (UID/MULTI/00612/2013) and the grant SFRH/BSAB/135473/2017.

REFERENCES

(1) Monthly Average Mauna Loa CO2

https://www.esrl.noaa.gov/gmd/ccgg/trends/index.html (accessed Jul 21, 2018).

(2) Centi, G.; Perathoner, S. Opportunities and Prospects in the Chemical Recycling of Carbon Dioxide to Fuels. Catal. Today 2009, 148, 191–205.

(3) Electrochemical Reduction of Carbon Dioxide - Overcoming the Limitations of Photosynthesis; Marken, F., Fermin, D., Eds.; Energy and Environment Series; Royal Society of Chemistry: Cambridge, 2018.

(4) Qiao, J.; Liu, Y.; Hong, F.; Zhang, J. A Review of Catalysts for the Electroreduction of Carbon Dioxide to Produce Low-Carbon Fuels. Chem. Soc. Rev. 2014, 43, 631–675.

(5) Angamuthu, R.; Byers, P.; Lutz, M.; Spek, A. L.; Bouwman, E. Electrocatalytic CO2 Conversion to Oxalate by a Copper Complex. Science 2010, 327, 313–315.

(6) Windle, C. D.; Perutz, R. N. Advances in Molecular Photocatalytic and Electrocatalytic CO2 Reduction. Coord. Chem. Rev. 2012, 256, 2562–2570.

(7) Hawecker, J.; Lehn, J.-M.; Ziessel, R. Electrocatalytic Reduction of Carbon Dioxide Mediated by Re(Bipy)(CO)3Cl (Bipy = 2,2ʹ-Bipyridine). J. Chem. Soc. Chem. Commun. 1984, 984, 328..

(8) Machan, C. W.; Chabolla, S. A.; Yin, J.; Gilson, M. K.; Tezcan, F. A.; Kubiak, C. P. Supramolecular Assembly Promotes the Electrocatalytic Reduction of Carbon Dioxide by Re(I) Bipyridine Catalysts at a Lower Overpotential. J. Am. Chem. Soc. 2014, 136, 14598–14607.

(9) Ching, H. Y. V.; Wang, X.; He, M.; Perujo Holland, N.; Guillot, R.; Slim, C.; Griveau, S.; Bertrand, H. C.; Policar, C.; Bedioui, F.; et al. Rhenium Complexes Based on 2-Pyridyl-1,2,3-Triazole Ligands: A New Class of CO2 Reduction Catalysts. Inorg. Chem. 2017, 56, 2966–2976.

(10) Wilting, A.; Stolper, T.; Mata, R. A.; Siewert, I. Dinuclear Rhenium Complex with a Proton Responsive Ligand as a Redox Catalyst for the Electrochemical CO2 Reduction. Inorg. Chem. 2017, 56, 4176–4185.

(11) Stanton, C. J.; Machan, C. W.; Vandezande, J. E.; Jin, T.; Majetich, G. F.; Schaefer, H. F.; Kubiak, C. P.; Li, G.; Agarwal, J. Re(I) NHC Complexes for Electrocatalytic Conversion of CO2. Inorg. Chem. 2016, 55, 3136–3144.

(12) Maurin, A.; Ng, C.-O.; Chen, L.; Lau, T.-C.; Robert, M.; Ko, C.-C. Photochemical and Electrochemical Catalytic Reduction of CO2 with NHC-Containing Dicarbonyl Rhenium(I) Bipyridine Complexes. Dalton Trans. 2016, 45, 14524–14529

(13) Hadadzadeh, H.; Farrokhpour, H.; Simpson, J.; Shakeri, J.; Daryanavard, M.; Shokrollahi, M. Electrocatalytic Reduction of CO2 to CO by a Mononuclear Ruthenium(II) Complex. New J. Chem. 2016, 40, 6347–6357.

(14) Chen, Z.; Chen, C.; Weinberg, D. R.; Kang, P.; Concepcion, J. J.; Harrison, D. P.; Brookhart, M. S.; Meyer, T. J. Electrocatalytic Reduction of CO2 to CO by Polypyridyl Ruthenium Complexes. Chem. Commun. 2011, 47, 12607-12609.

(15) Daryanavard, M.; Hadadzadeh, H.; Weil, M.; Farrokhpour, H. Electrocatalytic Reduction of CO2 to CO in the Presence of a Mononuclear Polypyridyl Ruthenium(II) Complex. J. CO2 Util. 2017, 17, 80–89.

Page 21: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

(16) Szymaszek, A.; Pruchnik, F. P. Electrochemical Reduction of Carbon Dioxide in the Presence of RhCl(CO)(PPh3)2 and IrCl(CO)(PPh3)2. J. Organomet. Chem. 1989, 376, 133–140.

(17) Ahn, S. T.; Bielinski, E. A.; Lane, E. M.; Chen, Y.; Bernskoetter, W. H.; Hazari, N.; Palmore, G. T. R. Enhanced CO2 Electroreduction Efficiency through Secondary Coordination Effects on a Pincer Iridium Catalyst. Chem. Commun. 2015, 51, 5947–5950.

(18) Sypaseuth, F. D.; Matlachowski, C.; Weber, M.; Schwalbe, M.; Tzschucke, C. C. Electrocatalytic Carbon Dioxide Reduction by Using Cationic Pentamethylcyclopentadienyl-Iridium Complexes with Unsymmetrically Substituted Bipyridine Ligands. Chem. Eur. J. 2015, 21, 6564–6571.

(19) Roy, S.; Sharma, B.; Pécaut, J.; Simon, P.; Fontecave, M.; Tran, P. D.; Derat, E.; Artero, V. Molecular Cobalt Complexes with Pendant Amines for Selective Electrocatalytic Reduction of Carbon Dioxide to Formic Acid. J. Am. Chem. Soc. 2017, 139, 3685–3696.

(20) Neri, G.; Walsh, J. J.; Wilson, C.; Reynal, A.; Lim, J. Y. C.; Li, X.; White, A. J. P.; Long, N. J.; Durrant, J. R.; Cowan, A. J. A Functionalised Nickel Cyclam Catalyst for CO2 reduction: Electrocatalysis, Semiconductor Surface Immobilisation and Light-Driven Electron Transfer. Phys. Chem. Chem. Phys. 2015, 17, 1562–1566.

(21) Song, J.; Klein, E. L.; Neese, F.; Ye, S. The Mechanism of Homogeneous CO2 reduction by Ni(Cyclam): Product Selectivity, Concerted Proton-Electron Transfer and C-O Bond Cleavage. Inorg. Chem. 2014, 53, 7500–7507.

(22) Bourrez, M.; Molton, F.; Chardon-Noblat, S.; Deronzier, A. [Mn(Bipyridyl)(CO)3Br]: An Abundant Metal Carbonyl Complex as Efficient Electrocatalyst for CO2 Reduction. Angew. Chemie Int. Ed. 2011, 50, 9903–9906.

(23) Spall, S. J. P.; Keane, T.; Tory, J.; Cocker, D. C.; Adams, H.; Fowler, H.; Meijer, A. J. H. M.; Hartl, F.; Weinstein, J. A. Manganese Tricarbonyl Complexes with Asymmetric 2-Iminopyridine Ligands: Toward Decoupling Steric and Electronic Factors in Electrocatalytic CO2 Reduction. Inorg. Chem. 2016, 55, 12568–12582.

(24) Sampson, M. D.; Kubiak, C. P. Manganese Electrocatalysts with Bulky Bipyridine Ligands: Utilizing Lewis Acids To Promote Carbon Dioxide Reduction at Low Overpotentials. J. Am. Chem. Soc. 2016, 138, 1386–1393.

(25) Sampson, M. D.; Nguyen, A. D.; Grice, K. a.; Moore, C. E.; Rheingold, A. L.; Kubiak, C. P. Manganese Catalysts with Bulky Bipyridine Ligands for the Electrocatalytic Reduction of Carbon Dioxide: Eliminating Dimerization and Altering Catalysis. J. Am. Chem. Soc. 2014, 136, 5460–5471.

(26) Smieja, J. M.; Sampson, M. D.; Grice, K. A.; Benson, E. E.; Froehlich, J. D.; Kubiak, C. P. Manganese as a Substitute for Rhenium in CO2 Reduction Catalysts: The Importance of Acids. Inorg. Chem. 2013, 52, 2484–2491.

(27) Franco, F.; Cometto, C.; Nencini, L.; Barolo, C.; Sordello, F.; Minero, C.; Fiedler, J.; Robert, M.; Gobetto, R.; Nervi, C. Local Proton Source in Electrocatalytic CO2 Reduction with [Mn(Bpy-R)(CO)3Br] Complexes. Chem. Eur. J. 2017, 23, 4782–4793.

(28) Grills, D. C.; Ertem, M. Z.; McKinnon, M.; Ngo, K. T.; Rochford, J. Mechanistic Aspects of CO2 Reduction Catalysis with Manganese-Based Molecular Catalysts. Coord. Chem. Rev. 2018, 374, 173–217.

(29) Azcarate, I.; Costentin, C.; Robert, M.; Savéant, J. M. Through-Space Charge Interaction Substituent Effects in Molecular Catalysis Leading to the Design of the Most Efficient Catalyst of CO2-to-CO Electrochemical Conversion. J. Am. Chem. Soc. 2016, 138, 16639–16644.

(30) Costentin, C.; Robert, M.; Savéant, J. M. Current Issues in Molecular Catalysis Illustrated by Iron Porphyrins as Catalysts of the CO2-to-CO Electrochemical Conversion. Acc. Chem. Res. 2015, 48, 2996–3006.

(31) Mohamed, E. A.; Zahran, Z. N.; Naruta, Y. Efficient Electrocatalytic CO2 Reduction with a Molecular Cofacial Iron Porphyrin Dimer. Chem. Commun. 2015, 51, 16900–16903.

(32) Elgrishi, N.; Chambers, M. B.; Wang, X.; Fontecave, M. Molecular Polypyridine-Based Metal Complexes as Catalysts for the Reduction of CO2. Chem. Soc. Rev. 2017, 46, 761–796.

(33) Francke, R.; Schille, B.; Roemelt, M. Homogeneously Catalyzed Electroreduction of Carbon Dioxide - Methods, Mechanisms, and Catalysts. Chem. Rev. 2018, 118, 4631–4701.

(34) Grice, K. A. Carbon Dioxide Reduction with Homogenous Early Transition Metal Complexes: Opportunities and Challenges for Developing CO2 Catalysis. Coord. Chem. Rev. 2017, 336, 78–95.

(35) Grice, K. A.; Saucedo, C. Electrocatalytic Reduction of CO2 by Group 6 M(CO)6 Species without “Non-Innocent” Ligands. Inorg. Chem. 2016, 55, 6240–6246.

(36) Clark, M. L.; Grice, K. A.; Moore, C. E.; Rheingold, A. L.; Kubiak, C. P. Electrocatalytic CO Reduction by M(Bpy-R)(CO)4 (M = Mo, W; R = H, tBu) Complexes. Electrochemical, Spectroscopic, and Computational Studies and Comparison with Group 7 Catalysts. Chem. Sci. 2014, 5, 1894–1900.

(37) Tory, J.; Setterfield-Price, B.; Dryfe, R. A. W.; Hartl, F. [M(CO)4(2,2′-Bipyridine)] (M= Cr, Mo, W) Complexes as Efficient Catalysts for Electrochemical Reduction of CO2 at a Gold Electrode. ChemElectroChem 2015, 2, 213–217.

(38) Franco, F.; Cometto, C.; Sordello, F.; Minero, C.; Nencini, L.; Fiedler, J.; Gobetto, R.; Nervi, C. Electrochemical Reduction of CO2 by M(CO)4(Diimine) Complexes (M= Mo, W): Catalytic Activity Improved by 2,2′-Dipyridylamine. ChemElectroChem 2015, 2, 1372–1379.

(39) Hartl, F.; Taylor, J. O.; Leavey, R. D. Solvent and Ligand Substitution Effects on Electrocatalytic Reduction of CO2 with [Mo(CO)4(x,xʹ-dimethyl-2,2′-bipyridine)] (x = 4-6) Enhanced at a Gold Cathodic Surface. ChemElectroChem. DOI: 10.1002/celc.201800879

(40) Sieh, D.; Lacy, D. C.; Peters, J. C.; Kubiak, C. P. Reduction of CO2 by Pyridine Monoimine Molybdenum Carbonyl Complexes: Cooperative Metal-Ligand Binding of CO2. Chem. Eur. J. 2015, 21, 8497–8503.

(41) Ngo, K. T.; McKinnon, M.; Mahanti, B.; Narayanan, R.; Grills, D. C.; Ertem, M. Z.; Rochford, J. Turning on the Protonation-First Pathway for Electrocatalytic CO2 Reduction by Manganese Bipyridyl Tricarbonyl Complexes. J. Am. Chem. Soc. 2017, 139, 2604–2618.

(42) a) Neri, G.; Donaldson, P. M.; Cowan, A. J. The Role of Electrode–Catalyst Interactions in Enabling Efficient CO2 Reduction with Mo(Bpy)(CO)4 As Revealed by Vibrational Sum-Frequency Generation Spectroscopy. J. Am. Chem. Soc. 2017, 139, 13791–13797. b) Neri, G.; Teobaldi, G.; Walsh, J. J.; Donaldson, P.M. & Cowan A. J. Detection of catalytic intermediates at an electrode surface during carbon dioxide reduction by an earth-abundant catalyst. Nat. Catal. Accepted.

Page 22: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

(43) Rotundo, L.; Garino, C.; Gobetto, R.; Nervi, C. Computational Study of the Electrochemical Reduction of W(CO)4(2,2′-Dipyridylamine). Inorg. Chim. Acta 2018, 470, 373–378.

(44) Ryan, D. E.; Cardin, D. J.; Hartl, F. η3-Allyl Carbonyl Complexes of Group 6 Metals: Structural Aspects, Isomerism, Dynamic Behaviour and Reactivity. Coord. Chem. Rev. 2017, 335, 103–149.

(45) Tory, J.; Gobaille-Shaw, G.; Chippindale, A. M.; Hartl, F. Spectroelectrochemical Study of Complexes [Mo(CO)2(η3-allyl)(α-Diimine)(NCS)] (α-Diimine = Bis(2,6-Dimethylphenyl)-Acenaphthenequinonediimine and 2,2′-Bipyridine) Exhibiting Different Molecular Structure and Redox Reactivity J. Organomet. Chem. 2014, 760, 30–41.

(46) Goodyear, J. W.; Hemingway, C. W.; Harrington, R. W.; Wiseman, M. R.; Brisdon, B. J. The Crystal Structure of [Mo(NCS)(CO)2(η3-C3H5)(NCMe)2]·MeCN and the Reactions of {Mo(CO)2(η3-C3H5)+} Containing Species with Symmetric Alkynes. J. Organomet. Chem. 2002, 664, 176–181.

(47) Agilent (2014). CrysAlis PRO. Agilent Technologies Ltd, Yarnton, Oxfordshire, England.

(48) Palatinus, L.; Chapuis, G. SUPERFLIP – a Computer Program for the Solution of Crystal Structures by Charge Flipping in Arbitrary Dimensions. J. Appl. Crystallogr. 2007, 40, 786–790.

(49) Betteridge, P. W.; Carruthers, J. R.; Cooper, R. I.; Prout, K.; Watkin, D. J.; IUCr. CRYSTALS Version 12: Software for Guided Crystal Structure Analysis. J. Appl. Crystallogr. 2003, 36, 1487–1487.

(50) van der Sluis, P.; Spek, A. L. BYPASS: An Effective Method for the Refinement of Crystal Structures Containing Disordered Solvent Regions. Acta Crystallogr. Sect. A 1990, 46, 194–201.

(51) Krejčik, M.; Daněk, M.; Hartl, F. Simple Construction of an Infrared Optically Transparent Thin-Layer Electrochemical Cell: Applications to the Redox Reactions of ferrocene, Mn2(CO)10 and Mn(CO)3(3,5-di-t-Butyl-Catecholate)-. J. Electroanal. Chem. Interfacial Electrochem. 1991, 317, 179–187.

(52) Mahabiersing, T.; Lyuten, H.; Nieuwendam, R. C.; Hartl, F. Synthesis, Spectroscopy and Spectroelectrochemistry of Chlorocarbonyl {1,2-Bis[(2,6-Diisopropylphenyl)imino]ace-naphthene-κ2-N,Nʹ}rhodium(I). Collect. Czech. Chem. Commun. 2003, 68, 1687–1709.

(53) Parr, R. G.; Yang, W. Density-Functional Theory of Atoms and Molecules; Oxford University Press, 1989.

(54) te Velde, G.; Bickelhaupt, F. M.; Baerends, E. J.; Fonseca Guerra, C.; van Gisbergen, S. J. A.; Snijders, J. G.; Ziegler, T. Chemistry with ADF. J. Comput. Chem. 2001, 22, 931–967.

(55) Fonseca Guerra, C.; Snijders, J. G.; Te Velde, G.; Baerends, E. J. Towards an Order-N DFT Method. Theor. Chem. Acc. 1998, 99, 391–403.

(56) Theoretical Chemistry, Vrije Universiteit, Amsterdam, T. N. ADF2013 SCM http://www.scm.com (accessed Jul 26, 2018).

(57) Vosko, S. H.; Wilk, L.; Nusair, M. Accurate Spin-Dependent Electron Liquid Correlation Energies for Local Spin Density Calculations: A Critical Analysis. Can. J. Phys. 1980, 58, 1200–1211.

(58) Becke, A. D. A New Inhomogeneity Parameter in Density-Functional Theory. J. Chem. Phys. 1998, 109, 2092.

(59) Perdew, J. P. Density-Functional Approximation for the Correlation Energy of the Inhomogeneous Electron Gas. Phys. Rev. B 1986, 33, 8822–8824.

(60) Perdew, J. P. Erratum: Density-Functional Approximation for the Correlation Energy of the Inhomogeneous Electron Gas. Phys. Rev. B 1986, 34, 7406–7406.

(61) van Lenthe, E.; Ehlers, A.; Baerends, E.-J. Geometry Optimizations in the Zero Order Regular Approximation for Relativistic Effects. J. Chem. Phys. 1999, 110, 8943-8953.

(62) Moussa, J.; Chamoreau, L.-M.; Degli Esposti, A.; Gullo, M. P.; Barbieri, A.; Amouri, H. Tuning Excited States of Bipyridyl Platinum(II) Chromophores with π-Bonded Catecholate Organometallic Ligands: Synthesis, Structures, TD-DFT Calculations, and Photophysical Properties. Inorg. Chem. 2014, 53, 6624–6633.

(63) van Gisbergen, S. J. A.; Rosa, A.; Ricciardi, G.; Baerends, E. J. Time-Dependent Density Functional Calculations on the Electronic Absorption Spectrum of Free Base Porphin. J. Chem. Phys. 1999, 111, 2499-2506.

(64) Rosa, A.; Baerends, E. J.; van Gisbergen, S. J. A.; van Lenthe, E.; Groeneveld, J. A.; Snijders, J. G. Electronic Spectra of M(CO)6 (M = Cr, Mo, W) Revisited by a Relativistic TDDFT Approach. J. Am. Chem. Soc. 1999, 121, 10356–10365.

(65) van Gisbergen, S. J. A.; Groeneveld, J. A.; Rosa, A.; Snijders, J. G.; Baerends, E. J. Excitation Energies for Transition Metal Compounds Applications to MnO4

-, Ni(CO)4, and Mn2(CO)10. J. Phys. Chem. A 1999, 103, 6835–6844.

(66) Wang, F.; Ziegler, T. A Simplified Relativistic Time-Dependent Density-Functional Theory Formalism for the Calculations of Excitation Energies Including Spin-Orbit Coupling Effect. J. Chem. Phys. 2005, 123, 154102.

(67) Ziegler, T.; Rauk, A. Carbon Monoxide, Carbon Monosulfide, Molecular Nitrogen, Phosphorus Trifluoride, and Methyl Isocyanide as Sigma Donors and Pi Acceptors. A Theoretical Study by the Hartree-Fock-Slater Transition-State Method. Inorg. Chem. 1979, 18, 1755–1759.

(68) Ziegler, T.; Rauk, A. On the Calculation of Bonding Energies by the Hartree Fock Slater Method. Theor. Chim. Acta 1977, 46, 1–10.

(69) Chemcraft Program http://www.chemcraftprog.com/index.html (accessed Jul 26, 2018).

(70) Graham, A. J.; Fenn, R. H. The Crystal Structure of Some Dicarbonyl-Allyl Derivatives of Substituted Carbonyls of Molybdenum (I). The Crystal Structure of (Isothiocyanato)dicarbonyl-2,2′-bipyridine-π-allylmolybdenum NCS(CO)2(C10H8N2)(π-C3H5)Mo. J. Organomet. Chem. 1969, 17, 405–422.

(71) Liu, F. C.; Yang, P. S.; Chen, C. Y.; Lee, G. H.; Peng, S. M. Syntheses, Structures, and Dynamic Properties of M(CO)2(η3-C3H5)(L-L)(NCBH3) (M = Mo, W; L-L = Dppe, Bipy, En). J. Organomet. Chem. 2008, 693, 537–545.

(72) Ascenso, J. R.; De Azevedo, C. G.; Calhorda, M. J.; De, M. A. A. F.; Costa, P.; Dias, A. R.; Drew, M. G. B.; Félix, V.; Galvão, A. M.; Romão, C. C. Synthesis, Bonding and Dynamic Behavior of Fac-[Mo(II)(CO)2(η3-Allyl)] Derivatives. J. Organomet. Chem. 2001, 632, 197–208.

(73) Faller, J. W.; Haitko, D. A.; Adams, R. D.; Chodosh, D. F. Aspects of Chirality Retention in Rearrangements of Pseudooctahedral

Page 23: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

Molybdenum and Tungsten Complexes. J. Am. Chem. Soc. 1979, 101, 865–876.

(74) Alonso, J. C.; Neves, P.; Da Silva, M. J. P.; Quintal, S.; Vaz, P. D.; Silva, C.; Valente, A. A.; Ferreira, P.; Calhorda, M. J.; Félix, V.; et al. Molybdenum η3-Allyl Dicarbonyl Complexes as a New Class of Precursors for Highly Reactive Epoxidation Catalysts with Tert-Butyl Hydroperoxide. Organometallics 2007, 26, 5548–5556.

(75) Chabolla, S. A.; Dellamary, E. A.; Machan, C. W.; Tezcan, F. A.; Kubiak, C. P. Combined Steric and Electronic Effects of Positional Substitution on Dimethyl-Bipyridine Rhenium(I) Tricarbonyl Electrocatalysts for the Reduction of CO2. Inorganica Chim. Acta 2014, 422, 109–113.

(76) Vichova, J.; Hartl, F.; Vlcek, A. Wavelength-Dependent Photosubstitution and Excited-State Dynamics of [Cr(CO)4(2,2ʹ-Bipyridine)]: A Quantum Yield and Picosecond Absorption Study. J. Am. Chem. Soc. 1992, 114, 10903–10910.

(77) Hartl, F.; Rosa, P.; Ricard, L.; Le Floch, P.; Záliš, S. Electronic Transitions and Bonding Properties in a Series of Five-Coordinate “16-Electron” Complexes [Mn(CO)3(L2)]- (L2 = Chelating Redox-Active π-Donor Ligand). Coord. Chem. Rev. 2007, 251, 557–576.

(78) van Outersterp, J. W. M.; Hartl, F.; Stufkens, D. J. Variable Temperature IR Spectroelectrochemical Investigation of the Stability of the Metal-Metal-Bonded Radical Anions [(CO)5MnRe(CO)3(L)]•- (L = 2,2ʹ-Bipyridine (BPY), 2,2ʹ-

Bipyrimidine (BPYM), 2,3-Bis(2-Pyridyl)Pyrazine (DPP)) and [(CO)5MnRe(CO)3(L)Re(Br)(CO)3]- (L = BPYM, DPP) Controlled by the Lowest π* (α-Diimine) Orbital Energy. Organometallics 1995, 14, 3303–3310.

(79) An alternative structure of 6-coordinate [1-PrCN]‒ has been sug-gested by DFT calculations on its possible triplet states. The para-

magnetic equatorial form appears to be quite stable, revealing a changed η2-allyl coordination and a stronger Mo‒PrCN bond. A

detailed investigation of this interesting alternative was out of the

scope of this work, but attention will be paid to it in the follow-up

studies.

(80) Stor, G. J.; Hartl, F.; van Outersterp, J. W. M.; Stufkens, D. J. Spectroelectrochemical (IR, UV/Vis) Determination of the Reduction Pathways for a Series of [Re(CO)3(α-Diimine)Lʹ]0/+

(Lʹ = Halide, Otf‒, THF, MeCN, n-PrCN, PPh3, P(OMe)3) Complexes. Organometallics 1995, 14, 1115–1131.

(81) Hartl, F.; Aarnts, M. P.; Nieuwenhuis, H. A.; Van Slageren, J. Electrochemistry of Different Types of Photoreactive Ruthenium(II) Dicarbonyl α-Diimine Complexes. Coord. Chem. Rev. 2002, 230, 107–125.

(82) Zeng, Q.; Tory, J.; Hartl, F. Electrocatalytic Reduction of Carbon Dioxide with a Manganese (I) Tricarbonyl Complex Containing a Nonaromatic α-Diimine Ligand. Organometallics 2014, 33, 5002–5008

Page 24: Group 6 metal complexes as electrocatalysts of CO2 ...centaur.reading.ac.uk/81186/1/Main text revised 25... · Preliminary CV and IR SEC results reveal differences in the strength

Table of Contents artwork